Science + Tech – The Conversation

Does the exodus to UpScrolled signify the end of TikTok?

2026-02-04T16:34:28Z

Soon after American investors took control of TikTok's U.S. operations, users started complaining that content on certain topics was being suppressed. (Unsplash/Appshunter.io)

Until recently, you might have never heard of the TikTok competitor UpScrolled. But as of Jan. 29, the app reached No. 1 one in Apple’s app store as disgruntled TikTok users in the United States rushed to sign up.

The exodus to UpScrolled comes after a group of American investors, including Oracle founder Larry Ellison, acquired a majority stake in TikTok’s U.S. operations on Jan. 22, a day before the deadline set by President Donald Trump for the app’s U.S operations to be separated from Chinese parent company ByteDance.

Trump and other American officials have long pushed for acquiring TikTok’s U.S. operations, citing concerns over China accessing the data of U.S. citizens. However, soon after the acquisition, TikTok users started complaining of shadow banning, a disputed tactic whereby people suggest social media sites will allow you to post, but will not allow anyone else to see what you post.

The acquisition comes amid civil unrest in the U.S. as Immigration and Customs Enforcement officers (ICE) conduct raids in cities like Minneapolis that have resulted in multiple deaths and hospitalizations. Concerned users have been uploading video documenting ICE’s actions, but began to notice their videos were not garnering any attention on TikTok, or sometimes, not uploading at all following the acquisition.

Users posting about other topics such as Palestine have also expressed concerns about censorship. Palestinian journalist Bisan Owda’s account was banned shortly after the acquisition. It was restored following an outcry from users.

TikTok says anyone experiencing a disruption over the last couple of weeks has not been shadow banned; it was result of technical problems following a polar vortex and associated weather-related issues. But this statement from U.S. TikTok came after one million downloads of UpScrolled and reports of concerned users deleting TikTok.

Controlling the algorithm

It may indeed be a coincidence that people had trouble uploading videos critical of ICE at a time of changing ownership, but the whole incident had users talking.

As part of the acquisition, TikTok has been programmed with a new U.S.-specific content moderation algorithm that influences what people do and don’t see. Like with every other social network, the algorithm is considered proprietary information, meaning no academic nor policymaker can independently audit it.

Trump has expressed interest in controlling social media algorithms, so it’s no wonder people are connecting the outage with possible censorship. Looking at Reddit posts about the TikTok sale reveals how upset some users are.

It’s well known that China engages in censorship on the Chinese version of TikTok, Douyin. In fact, this practice was commented on by France’s President Emmanuel Macron, who stated that children on TikTok in China receive more educational content than children in France do.

Knowing this, it’s not surprising that American users would connect the dots and suggest that any TikTok outage would be a result of government censorship.

The truth is, there’s no way to know for sure whether censorship did occur in the first week of the takeover, or whether it’s still occurring in less obvious ways now. Regardless of whether direct government interference is an issue, the algorithm still filters content in ways that often lead to misinformation spreading among a global user base.

Is time up for TikTok?

Does the rush of users from TikTok to apps like UpScrolled spell hard times ahead for TikTok U.S.? We’ve been here before, and the apps that take a temporary hit usually bounce back. After Elon Musk took over Twitter and rebranded it X in 2022, many users, including high-profile celebrities and corporations, left the platform. However, engagement is still strong among people who identify as right wing and MAGA.

Every couple of years, it seems, news outlets publish articles about reasons to leave Facebook. But Facebook and X are still going strong. The fact that these sites survive the exodus of both high-profile and regular users is likely due to network effects.

Social media platforms become more valuable the more people are on them. Not only do they become more interesting when there are more people posting content, but people also want to be on platforms where their friends, family and favourite celebrities already are.

Network effects mean that unless UpScrolled continues its explosive growth, people are unlikely to continue to choose it over the more established TikTok. At best, we might see a Twitter/X effect, which is where TikTok will host more pro-U.S. government content creators and those people who want to follow them, and UpScrolled will host more critical content creators and their followers. This is basically what happened when many left-leaning users moved to BlueSky as an alternative to X.

Because each social network engages in or facilitates different types of content filtering, each provides a different kind of echo chamber that people self-select into or out of.

These echo chambers are a problem because they reinforce beliefs, even ones grounded in mis- and disinformation, and in turn create deeper more polarized divisions between people that are hard to escape from. Since young people report getting most of their news from social media sites, people concerned about algorithms have more than just government censorship to worry about.

Jaigris Hodson receives funding from the Social Sciences and Humanities Research Council of Canada (SSHRC)

The mental edge that separates elite athletes from the rest

2026-02-03T16:36:32Z

Elite sport often looks like a test of speed, strength and technical skill. Yet some of the most decisive moments in high-level competition unfold too quickly to be explained by physical ability alone.

Consider Canadian hockey superstar Connor McDavid’s overtime goal at the 4 Nations Face-Off against the United States last February. The puck was on his stick for only a fraction of a second, the other team’s defenders were closing in and he still somehow found the one opening no one else saw.

As professional hockey players return to the ice at the Milan-Cortina Olympics, Canadians can expect more moments like this. Increasingly, research suggests these moments are better understood not as just physical feats, but also as cognitive ones.

A growing body of research suggests a group of abilities known as perceptual-cognitive skills are key differentiators. This is the mental capacity to turn a blur of sights, sounds and movements into split-second decisions.

These skills allow elite athletes to scan a chaotic scene, pick out the right cues and act before anyone else sees the opportunity. In short, they don’t just move faster, but they also see smarter.

Connor McDavid Wins 4 Nations Face-Off For Canada In Overtime (Sportsnet)

How athletes manage visual chaos

One way researchers study these abilities is through a task known as multiple-object tracking, which involves keeping tabs on a handful of moving dots on a screen while ignoring the rest. Multiple-object tracking is a core method I use in my own research on visual attention and visual-motor co-ordination.

Multiple-object tracking taxes attention, working memory and the ability to suppress distractions. These are the same cognitive processes athletes rely on to read plays and anticipate movement in real time.

Unsurprisingly, elite athletes reliably outperform non-athletes on this task. After all, reading plays, tracking players and anticipating movement all depend on managing visual chaos.

There is, however, an important caveat. Excelling at multiple-object tracking will not suddenly enable someone to anticipate a play like McDavid or burst past a defender like Marie-Philip Poulin, captain of the Canadian women’s hockey team. Mastering one narrow skill doesn’t always transfer to real-world performance. Researchers often describe this limitation as the “curse of specificity.”

This limitation raises a deeper question about where athletes’ mental edge actually comes from. Are people with exceptional perceptual-cognitive abilities drawn to fast-paced sports, or do years of experience sharpen it over time?

Evidence suggests the answer is likely both.

Born with it or trained over time?

Elite athletes, radar operators and even action video game players — all groups that routinely track dynamic, rapidly changing scenes — consistently outperform novices on perceptual-cognitive tasks.

At the same time, they also tend to learn these tasks faster, pointing to the potential role of experience in refining these abilities.

What seems to distinguish elite performers is not necessarily that they take in more information, but that they extract the most relevant information faster. This efficiency may ease their mental load, allowing them to make smarter, faster decisions under pressure.

My research at McMaster University seeks to solve this puzzle by understanding the perceptual-cognitive skills that are key differentiators in sport, and how to best enhance them.

This uncertainty around how to best improve perceptual-cognitive skills is also why we should be cautious about so-called “brain training” programs that promise to boost focus, awareness or reaction time.

The marketing is often compelling, but the evidence for broad, real-world benefits is far less clear. The value of perceptual-cognitive training hasn’t been disproven, but it hasn’t been tested rigorously enough in real athletic settings to provide compelling evidence. To date, though, tasks that include a perceptual element such as multiple-object tracking show the most promise.

Training perceptual-cognitive skills

Researchers and practitioners still lack clear answers about the best ways to train perceptual-cognitive skills, or how to ensure that gains in one context carry over to another. This doesn’t mean cognitive training is futile, but it does mean we need to be precise and evidence-driven about how we approach it.

Research does, however, point to several factors that increase the likelihood of real-world transfer.

Training is more effective when it combines high cognitive and motor demands, requiring rapid decisions under physical pressure, rather than isolated mental drills. Exposure to diverse stimuli matters as well, as it results in a brain that can adapt, not just repeat. Finally, training environments that closely resemble the game itself are more likely to produce skills that persist beyond the training session.

The challenge now is translating these insights from the laboratory into practical training environments. Before investing heavily in new perceptual-cognitive training tools, coaches and athletes need to understand what’s genuinely effective and what’s just a high-tech placebo.

For now, this means treating perceptual-cognitive training as a complement to sport-specific training, not as a substitute. Insights will also come from closer collaborations between researchers, athletes and coaches.

There is however, support for incorporating perceptual-cognitive tasks as an assessment of “game sense” to inform scouting decisions.

The real secret to seeing the game differently, then, is not just bigger muscles or faster reflexes. It’s a sharper mind, and understanding how it works could change how we think about performance, both on and off the ice.

Mallory Terry does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Lessons from the sea: Nature shows us how to get ‘forever chemicals’ out of batteries

2026-02-03T15:21:42Z

As the world races to electrify everything from cars to cities, the demand for high-performance, long-lasting batteries is soaring. But the uncomfortable truth is this: many of the batteries powering our “green” technologies aren’t as green as we might think.

Most commercial batteries rely on fluorinated polymer binders to hold them together, such as polyvinylidene fluoride. These materials perform well — they’re chemically stable, resistant to heat and very durable. But they come with a hidden environmental price.

Fluorinated polymers are derived from fluorine-containing chemicals that don’t easily degrade, releasing persistent pollutants called PFAS (per- and polyfluoroalkyl substances) during their production and disposal. Once they enter the environment, PFAS can remain in water, soil and even human tissue for hundreds of years, earning them the nickname “forever chemicals.”

We’ve justified their use because they increase the lifespan and performance of batteries. But if the clean energy transition relies on materials that pollute, degrade ecosystems and persist in the environment for years, is it really sustainable?

As a graduate student, I spent years thinking about how to make batteries cleaner — not just in how they operate, but in how they’re made. That search led me somewhere unexpected: the ocean.

Why binders are important

Most commercial batteries rely on fluorinated polymer binders to hold them together. These materials perform well but come with an environmental cost. (Unsplash/CHUTTERSNAP)

Every rechargeable battery has three essential components: two electrodes separated by a liquid electrolyte that allows charged atoms (ions) to flow between them. When you charge a battery, the ions move from one electrode to the other, storing energy.

When you use the battery, the charged atoms flow back to their original side, releasing that stored energy to power your phone, car or the grid.

Each electrode is a mixture of three parts: an active material that stores and releases energy, a conductive additive that helps electrons move and a binder that holds everything together.

The binder acts like glue, keeping particles in place and preventing them from dissolving during use. Without it, a battery would be unable to hold a charge after only a few uses.

Lessons from the sea

Many marine organisms have evolved in remarkable ways to attach themselves to wet, slippery surfaces. Mussels, barnacles, sandcastle worms and octopuses produce natural adhesives to stick to rocks, ship hulls and coral in turbulent water — conditions that would defeat most synthetic glues.

For mussels, the secret lies in molecules called catechols. These molecules contain a unique amino acid in their sticky proteins that helps them form strong bonds with surfaces and hardens almost instantly when exposed to oxygen. This chemistry has already inspired synthetic adhesives used to seal wounds, repair tendons and create coatings that stick to metal or glass underwater.

Building on this idea, I began exploring a related molecule called gallol. Like catechol in mussels, gallol is used by marine plants and algae to cling to wet surfaces. Its chemical structure is very similar to catechol, but it contains an extra functional group that makes it even more adhesive and versatile. It can form multiple types of strong, durable and reversible bonds — properties that make it an excellent battery binder.

Mussels use molecules called catechols to stick to surfaces. (Unsplash/Manu Mateo)

A greener solution

Working with Prof. Dwight S. Seferos at the University of Toronto, we developed a polymer binder based on gallol chemistry and paired it with zinc, a safer and more abundant metal than lithium. Unlike lithium, zinc is non-flammable and easier to source sustainably, making it ideal for large-scale applications.

The results were remarkable. Our gallol-based zinc batteries maintained 52 per cent higher energy efficiency after 8,000 charge-discharge cycles compared to conventional batteries that use fluorinated binders. In practical terms, that means longer-lasting devices, fewer replacements and a smaller environmental footprint.

Our findings are proof that performance and sustainability can go hand-in-hand. Many in industry might still view “green” and “effective” as competing priorities, with sustainability an afterthought. That logic is backwards.

We can’t build a truly clean energy future using polluting materials. For too long, the battery industry has focused on performance at any cost, even if that cost includes toxic waste, hard-to-recycle materials and unsustainable and unethical mining practices. The next generation of technologies must be sustainable by design, built from sources are renewable, biodegradable and circular.

Nature has been running efficient, self-renewing systems for billions of years. Mussels, shellfish and seaweeds build materials that are strong, flexible and biodegradable. No waste and no forever chemicals. It’s time we started paying attention.

The ocean holds more than beauty and biodiversity; it may also hold the blueprint for the future of energy storage. But realizing that future requires a cultural shift in science, one that rewards innovation that heals, not just innovation that performs.

We don’t need to sacrifice progress to protect the planet. We just need to design with the planet in mind.

This research was supported by the National Sciences and Engineering Research Council of Canada, the Canadian Foundation for Innovation, and the Ontario Research Fund. Alicia M. Battaglia received funding from the Ontario Graduate Scholarship Program.

ChatGPT is in classrooms. How should educators now assess student learning?

2026-02-02T16:56:17Z

Generative artificial intelligence (GenAI) is now a reality in higher education, with students and professors integrating chatbots into teaching, learning and assessment. But this isn’t just a technical shift; it’s reshaping how students and educators learn and evaluate knowledge.

Our recent qualitative study with 28 educators across Canadian universities and colleges — from librarians to engineering professors — suggests that we have entered a watershed moment in education.

We must grapple with the question: What exactly should be assessed when human cognition can be augmented or simulated by an algorithm?

Research about AI and academic integrity

In our review of 15 years of research that engages how AI affects cheating in education, we found that AI is a double-edged sword for schools.

On one hand, AI tools like online translators and text generators have become so advanced that they can write just like humans. This makes it difficult for teachers to detect cheating. Additionally, these tools can sometimes present fake news as facts or repeat unfair social biases, such as racism and sexism, found in the data used to train them.

On the other hand, the studies we reviewed showed AI can be a legitimate assistant that can make learning more inclusive. For instance, AI can provide support for students with disabilities or help those who are learning an additional language.

Because it’s nearly impossible to block every AI tool, schools should not just focus on catching cheaters. Instead, schools and post-secondary institutions can update their policies and provide better training for both students and teachers. This helps everyone learn how to use technology responsibly while maintaining a high standard of academic integrity.

Participants in our study positioned themselves not as enforcers, but as stewards of learning with integrity.

Their focus was on distinguishing between assistance that supports learning and assistance that substitutes for it. They identified three skill areas where assessment boundaries currently fall: prompting, critical thinking and writing.

Prompting: A legitimate and assessable skill

Participants widely viewed prompting — the ability to formulate clear and purposeful instructions for a chatbot — as a skill they could assess. Effective prompting requires students to break down tasks, understand concepts and communicate precisely.

Several noted that unclear prompts often produce poor outputs, forcing students to reflect on what they are really asking.

Prompting was considered ethical only when used transparently, drawing on one’s own foundational knowledge. Without these conditions, educators feared prompting may drift into over-reliance or uncritical use of AI.

Critical thinking

Educators saw strong potential for AI to support assessing critical thinking. Because chatbots can generate text that sounds plausible but may contain errors, omissions or fabrications, students must evaluate accuracy, coherence and credibility. Participants reported using AI-generated summaries or arguments as prompts for critique, asking students to identify weaknesses or misleading claims.

These activities align with a broader need to prepare students for work in a future where assessing algorithmic information will be a routine task. Several educators argued it would be unethical not to teach students how to interrogate AI-generated content.

Writing: Where boundaries tighten

Writing was the most contested domain. Educators distinguished sharply between brainstorming, editing and composition:

• Brainstorming with AI was acceptable when used as a starting point, as long as students expressed their own ideas and did not substitute AI suggestions for their own thinking.

• Editing with AI (for example, grammar correction) was considered acceptable only after students had produced original text and could evaluate whether AI-generated revisions were appropriate. Although some see AI as a legitimate support for linguistic diversity, as well as helping to level the field for those with disabilities or those who speak English as an additional language, others fear a future of language standardization where the unique, authentic voice of the student is smoothed over by an algorithm.

• Having chatbots draft arguments or prose was implicitly rejected. Participants treated the generative phase of writing as a uniquely human cognitive process that needs to be done by students, not machines.

Educators also cautioned that heavy reliance on AI could tempt students to bypass the “productive struggle” inherent in writing, a struggle that is central to developing original thought.

Our research participants recognized that in a hybrid cognitive future, skills related to AI, together with critical thinking are essential skills for students to be ready for the workforce after graduation.

Living in the post-plagiarism era

The idea of co-writing with GenAI brings us into an post-plagiarism era where AI is integrated into into teaching, learning and communication in a way that challenges us to reconsider our assumptions about authorship and originality.

This does not mean that educators no longer care about plagiarism or academic integrity. Honesty will always be important. Rather, in a post-plagiarism context, we consider that humans and AI co-writing and co-creating does not automatically equate to plagiarism.

Today, AI is disrupting education and although we don’t yet have all the answers, it’s certain that AI is here to stay. Teaching students to co-create with AI is part of learning in a post-plagiarism world.

Design for a socially just future

Valid assessment in the age of AI requires clearly delineating which cognitive processes must remain human and which can be legitimately cognitively offloaded. To ensure higher education remains a space for ethical decision-making especially in terms of teaching, learning and assessment, we propose five design principles, based on our research:

1. Explicit expectations: The educator is responsible for making clear if and how GenAI can be used in a particular assignment. Students must know exactly when and how AI is a partner in their work. Ambiguity can lead to unintentional misconduct, as well as a breakdown in the student-educator relationship.

2. Process over product: By evaluating drafts, annotations and reflections, educators can assess the learning process, rather than just the output, or the product.

3. Design assessment tasks that require human judgment: Tasks requiring high-level evaluation, synthesis and critique of localized contexts are areas where human agency is still important.

4. Developing evaluative judgment: Educators must teach students to be critical consumers of GenAI, capable of identifying its limitations and biases.

5. Preserving student voice: Assessments should foreground how students know what they know, rather than what they know.

Preparing students for a hybrid cognitive future

Educators in this study sought ethical, practical ways to integrate GenAI into assessment. They argued that students must understand both the capabilities and the limitations of GenAI, particularly its tendency to generate errors, oversimplifications or misleading summaries.

In this sense, post-plagiarism is not about crisis, but about rethinking what it means to learn and demonstrate knowledge in a world where human cognition routinely interacts with digital systems.

Universities and colleges now face a choice. They can treat AI as a threat to be managed, or they can treat it as a catalyst for strengthening assessment, integrity and learning. The educators in our study favour the latter.

Sarah Elaine Eaton receives funding from the Social Sciences and Humanities Research Council of Canada and the University of Calgary.

Rahul Kumar has received funding from SSHRC in the past.

Robert Brennan receives funding from the Natural Sciences and Engineering Research Council of Canada (NSERC).

Beatriz Antonieta Moya Figueroa does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Artemis II: The first human mission to the moon in 54 years launches soon — with a Canadian on board

2026-01-29T12:10:25Z

The crew of the new NASA moon rocket Artemis II at the Kennedy Space Center, including Jeremy Hansen of the Canadian Space Agency, on the far right. From left: Reid Wiseman, Victor Glover and Christina Koch. (NASA)

It’s been 54 years since the last Apollo mission, and since then, humans have not ventured beyond low-Earth orbit. But that’s all about to change with next month’s launch of the Artemis II mission from the Kennedy Space Center in Florida.

NASA has announced that — due to a liquid hydrogen leak discovered during the wet dress rehearsal on Feb. 3 — the launch of Artemis II has been moved up to March 6.

It will be the first crewed flight of NASA’s Artemis program and the first time since 1972 that humans have ventured to the moon. Onboard is Canadian astronaut Jeremy Hansen, who will be the first non-American to fly to the moon and will make Canada only the second country in the world to send an astronaut into deep space.

NASA announced that due to a liquid hydrogen leak discovered during the wet dress rehearsal that they are postponing the launch of Artemis II to the March launch window, which begins March 6th.

I am a professor, an explorer and a planetary geologist. For the past 15 years, I have been helping to train Hansen and other astronauts in geology and planetary science. I am also a member of the Artemis III Science Team and the principal investigator for Canada’s first ever rover mission to the moon.

NASA’s Artemis II SLS rocket and Orion spacecraft secured to the mobile launcher at NASA’s Kennedy Space Center in Florida. (NASA)

What will the mission achieve?

NASA’s Artemis program, launched in 2017, has the ambitious goal to return humans to the moon and to establish a lunar base in preparation for sending humans to Mars. The first mission, Artemis I, launched in late 2022. Following some delays, Artemis II is scheduled for launch as early as a week from now.

Onboard will be Hansen, along with his three American crew-mates.

This is an incredibly exciting mission. Artemis II is the first time humans have launched on NASA’s huge SLS (Space Launch System) rocket, and the first time humans have flown in the Orion spacecraft.

SLS is the most powerful rocket NASA has ever built, with the capability to send more than 27 metric tonnes of payload — equipment, instruments, scientific experiments and cargo — to the moon. The Orion spacecraft sits at the very top and is the crew’s ride to the moon. The Artemis II crew named their Orion capsule Integrity, a word they say embodies trust, respect, candour and humility.

An infographic produced by NASA showing the different parts of the Orion spacecraft. (NASA)

What will Artemis II crew do in space?

Following launch, the crew will carry out tests of Integrity’s essential life-support systems: the water dispenser, firefighting equipment, and, of course, the toilet. Did you know there was no toilet on the Apollo missions? Instead, the crews used “relief tubes.”

If everything looks good, the Artemis II will ignite what’s known as the Interim Cryogenic Propulsion Stage — part of the SLS rocket still connected to Integrity — to elevate the spacecraft’s orbit. If things are still looking good, the Orion spacecraft and its four human travellers will spend 24 hours in a high-Earth orbit up to 70,000 kilometres away from the planet.

For comparison, the International Space Station orbits the Earth at a mere 400 kilometres.

Following a series of tests and checks, the crew will conduct one of the most critical stages of the mission: the Trans-Lunar Injection, or TLI. This is the crucial moment that changes a spacecraft from orbiting the Earth — where the option to quickly return home remains — to sending it on its way to the moon and into deep space.

The Artemis II mission’s 10-day ‘figure-eight’ trajectory. (NASA)

Once the Integrity is on its way to the moon after TLI, there is no turning back — at least, not without going to the moon first. That’s because Artemis II — like the early Apollo missions — enters what’s called a “free-return trajectory” after the TLI. What this means is that even if Integrity’s engines fail completely, the moon’s gravity will naturally loop the spacecraft around it and aim it towards Earth.

After the three-day journey to the moon, the crew will carry out perhaps the most exciting stage of the mission: lunar fly-by. Integrity will loop around the far side of the moon, passing anywhere from 6,000 to 10,000 kilometres above its surface — much farther than any Apollo mission.

To quote Star Trek, at that most distant point, the Artemis II crew will have boldly gone where no (hu)man has gone before. This will be, quite literally, the farthest from Earth that any human being has ever travelled.

International effort to explore the moon

That a Canadian astronaut is part of the crew of Artemis II is a testament to the collaborative international nature of the Artemis program.

While NASA created the program and is the driving force, there are now 60 countries that have signed the Artemis Accords.

On Jan. 26, 2026, Oman became the 61st nation to sign the Artemis Accords. (NASA)

The foundation for the Artemis Accords is the recognition that international co-operation in space is intended not only to bolster space exploration but to enhance peaceful relationships among nations. This is particularly necessary now — perhaps more than any other time since the Cold War.

I truly hope that as Integrity returns from the moon’s far side, people around the world will pause — at least for a few moments — and be united in thinking of a better future. As American astronaut Bill Anders, who flew the first crewed Apollo mission to the moon, once said:

“We came all this way to explore the moon, and the most important thing is that we discovered the Earth.”

This is an updated story following NASA’s Feb. 3 announcement that it’s postponing the Artemis II aunch until the March launch window, which begins March 6.

Gordon Osinski founded the company Interplanetary Exploration Odyssey Inc. He receives funding from the Natural Sciences and Engineering Research Council of Canada and the Canadian Space Agency.

Why Iran keeps turning off the internet during mass protests

2026-01-28T15:29:26Z

What began on Dec. 28 in Tehran as a revolt against economic hardship and the collapse of the national currency quickly spread across dozens of other Iranian cities and provinces. People from diverse socioeconomic, religious and ethnic backgrounds joined what has become the largest anti-regime protest since the 1979 revolution.

Chants of “death to the dictator” and “death to Khamenei” echoed far beyond Tehran’s Grand Bazaar. As a response, the government shut off all internet services, leaving roughly 92 million Iranians in a digital blackout since Jan. 8.

The protests are not an isolated eruption but the latest chapter in a continuous cycle of uprisings from the 1999 student movement, the Green movement of 2009, the protests of 2017 and the bloody November of 2019, the “uprising of the thirsty” in 2021 and the Woman, Life, Freedom uprising of 2022. Each was driven by different grievances but united by a deepening crisis of legitimacy and governance.

For authoritarian regimes, internet blackouts are a powerful political tool of repression that conceal state violence.

Violence justified for ‘security’

As the protests spread, the regime responded by unleashing lethal violence on the streets. Security forces fired live ammunition and pellet guns at demonstrators, deployed tear gas, carried out mass arrests and raided medical facilities where injured protesters were being treated, including hospitals in Illam and Tehran.

Arrests have surpassed 40,000, while estimates of the death toll vary widely, with reports suggesting that tens of thousands have been killed during the most intense days of repression. In cities such as Rasht, witnesses documented massacres as protesters attempted to flee security forces.

At the same time, state media outlets and senior political and judicial officials labelled protesters “terrorist agents” serving the United States and Israel, rhetoric that helped legitimize extreme violence in the name of national security.

The internet blackout as political strategy

Plunging millions of people into digital darkness was not a security precaution but a deliberate strategy used to disrupt collective action, prevent the documentation of state violence and control what both domestic and international audiences could see.

Mobile data, broadband connections and even phone lines were cut across the country, leaving families unable to contact loved ones, protesters cut off from one another and the outside world largely blind to events inside Iran. This was neither an unprecedented move nor a temporary security response. Iranian authorities have repeatedly restricted or disabled internet and telephone access during periods of sociopolitical unrest.

Under blackout conditions, the internet is not simply a space for expression, it is vital infrastructure that allows for information to flow.

By fragmenting connectivity, the state does not need to erase every image or silence every voice. It only needs to prevent a shared public record from forming. Violence becomes harder to document, deaths harder to count and accountability easier to evade.

Diaspora activism under blackout conditions

Outside Iran, this enforced silence prompted a wave of digital mobilization.

Iranians in the diaspora and their allies turned to platforms such as X and Instagram, circulating the hashtag #DigitalBlackoutIran to draw global attention to the shutdown and the escalating repression inside Iran. The hashtag became a way to make absence visible, revealing that the lack of images, videos and updates was itself the product of deliberate regime suppression and crackdown.

As the blackout continues, what’s at stake is not simply connectivity but the ability to bear witness. The struggle over internet access in Iran is therefore a deeply political one: it’s a struggle over who’s allowed to narrate, who’s allowed to be seen and whose suffering is allowed to register as real.

This use of #DigitalBlackoutIran didn’t emerge in vaccuum. It drew on previous movements and uprisings in Iran, where independent journalists are tightly restricted and repressed, public dissent is criminalized and uprisings are often followed by violent crackdowns and information blackouts.

When people cannot safely gather, publish or speak openly, and when documentation is actively disrupted, hashtags become a way of speaking out and of preserving what might otherwise disappear.

They allow dispersed users to find one another and construct a shared narrative of what’s happening. In this sense, hashtags function as a tool for mobilization and advocacy and as living archives of protest, keeping a record of repression and resistance alive when the state seeks to fragment, deny or erase it.

Yet the very visibility that gives hashtag activism its power also makes it vulnerable under authoritarian rule.

In Iran, the regime does not rely solely on blocking platforms or cutting access. It also actively manipulates online conversations from within. Alongside internet shutdowns, blocking social media platforms and filtering news websites, the state deploys co-ordinated networks of pro-regime accounts, often referred to as a “cyber army,” to disrupt protest hashtags.

These accounts flood hashtags with abusive and degrading language, disinformation and conspiracy narratives. The aim is to make participation emotionally, psychologically and socially costly.

This strategy reflects a broader shift in how autocratic regimes manage dissent online. Rather than silencing opposition, they increasingly seek to dominate digital spaces by overwhelming them, blurring truth with falsehood, intimidation with debate and visibility with noise.

The communications blackout and the disruption of online space point to the same reality in Iran: both operate as deliberate strategies of repression embedded in the regime’s broader architecture of control and discipline.

Under these conditions, the role of Iranians in the diaspora, along with sustained international media coverage, becomes critical not only in countering the silencing of dissent within Iran, but also in resisting the systematic erasure, distortion and fragmentation of the country’s ongoing history of defiance.

Niloofar Hooman does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Scientists have identified unique sounds for 8 fish species

2026-01-28T13:27:07Z

Have you ever wished you could swim like a fish? How about speak like one?

In a paper recently published in the Journal of Fish Biology, our team from the University of Victoria deciphered some of the strange and unique sounds made by different fish species along the coast of British Columbia.

Researchers have known for centuries that some fish make sounds, and the ancient Greek philosopher and scientist Aristotle even mentioned fish sounds in his writings. However, our understanding of which sounds are made by which fish species is extremely limited because it is difficult to pinpoint where a sound comes from underwater.

To accurately identify which sound is made by which fish, our team deployed an underwater acoustic localization array at sites in Barkley Sound, B.C. The localization array was designed by our project collaborator, Xavier Mouy, and it allowed us to precisely triangulate sounds to specific co-ordinates.

Using this triangulation and paired underwater video recordings, we were able to tie fish sounds to the correct species. We identified more than 1,000 fish sounds during our study, and successfully tied those sounds to eight different rocky reef fish species: copper, quillback, black, canary and vermillion rockfish, as well as lingcod, pile perch and kelp greenling.

We were particularly excited to identify sounds for canary and vermillion rockfish since these species had never been documented making sounds.

Differentiating fish sounds

We also wanted to investigate if different species sounds were unique enough to be differentiated from each other. We created a machine learning model using 47 different sound characteristics, like frequency (how high- or low-pitched the sound is) and duration (how long the sound is), to understand the unique differences in species calls.

For example, black rockfish make a long, growling sound similar to a frog croak, and quillback rockfish make a series of short knocks and grunts. The fish sound model was able to predict which sounds belonged to which species with up to 88 per cent accuracy. This was surprising and exciting to our team since many rocky reef fish species are very closely related.

Some fish species are known to make unique sounds during specific activities like courtship or guarding territory. Our research found that many species are also making sounds while fleeing from other fish.

For example, the copper and quillback rockfish both make significantly more grunting type sounds while being pursued by larger fish. We also documented sounds made during feeding activities and during aggressive activities like chasing.

Using sounds in future research

We also used stereo cameras in our research which allowed us to measure the length of the fish. We found that smaller fish make higher frequency (pitched) sounds than larger fish, which means scientists may eventually be able to estimate how big a fish is just by listening to its sounds. This discovery could be used in conservation in the future because estimating fish size is an important tool for effectively managing fish populations.

Our team plans to apply this research to improve marine conservation efforts. Now that we understand fish species sounds can be differentiated, there are many exciting possibilities for developing these acoustic tools into monitoring methods.

We can create species-specific fish sound detectors that will tell us where fish live without disturbing them. This has important implications for future conservation efforts, and the techniques we used can be adapted by scientists all over the world to decipher other fish calls.

Going forward, our team plans to develop a method of counting fish using acoustic recordings by examining the number of calls each species makes.

We also plan to compare the fish sounds we collected in Barkley Sound to fish calls made in other areas of British Columbia to see if fish have unique accents or dialects.

Using underwater sound recordings to study fish is highly beneficial. It is minimally invasive and acoustic recorders can collect information for months or years in hard to access or low visibility locations underwater. With more development, underwater acoustic monitoring could become an important new tool for conservationists and fisheries managers.

Darienne Lancaster has received funding from the Natural Sciences and Engineering Research Council of Canada (NSERC) and Fisheries and Oceans Canada Competitive Science Research Fund (CSRF). She is affiliated with Fisheries and Oceans Canada.

The invisible bubbles that spread cancer could also help stop it

2026-01-26T13:34:31Z

Cancer is transported from one organ to another by invisible bubbles. Understanding these microscopic messengers could change the fight against metastasis.

Preventing cancer from spreading throughout the body is the goal of our team at the Department of Electrical Engineering at the École de technologie supérieure (ÉTS). In collaboration with Prof. Julia Burnier and biology specialists at the Research Institute of the McGill University Health Centre, we are working to understand how cancers transform into metastases; in other words, how they invade other organs.

For about eight years, my team has been studying lipid nanoparticles, which are barely 100 nanometres in size and invisible to the naked eye. Our first task is to understand the path of metastasis. Then we try to determine different ways to inject drugs into the body.

Lipid nanoparticles, such as liposomes, differ from conventional approaches to cancer treatment because they deliver drugs directly to tumour cells. That increases their effectiveness and reduces toxicity compared to conventional chemotherapy.

Researchers have demonstrated that liposomes target tumours more effectively and reduce side effects, while others have observed that these nanomedicines improve the penetration and specificity of treatment, particularly in the case of metastases. These results confirm that nanomedicines can make cancer treatments more targeted, more effective and improve their tolerability.

Tiny particles responsible for spreading

Every cell in our body, whether healthy or cancerous, releases tiny particles called extracellular vesicles. These small bubbles, made of lipids and proteins, also carry genetic information.

When cancer cells release vesicles into the bloodstream, these vesicles can influence distant organs by modifying the behaviour of normal cells. This may create a more favourable environment for cancer cells to later settle and grow — a concept thought to contribute to metastasis.

The problem is, extracting and studying these natural vesicles is a long and difficult process. To speed up our research, my team produces artificial copies called liposomes using small devices called micromixers. By mixing different solutions — lipids, proteins, water and ethanol — our research team creates particles that resemble natural vesicles. The challenge is then to understand which lipids and proteins are contained in extracellular vesicles in order to produce liposomes.

We then inject these liposomes into liver cancer cells to see how they react. The more the cells retain these particles, the more it proves that the copies mimic reality well.

In a typical experiment, liposomes are manufactured with precise parameters to reproduce the size and charge of extracellular vesicles. They are also made visible by staining them with a fluorescent marker.

These liposomes are then incubated with cancer cells grown in our laboratory. This makes it possible to film and measure, in real time and without disturbing the cells, how and at what speed the liposomes are absorbed and expressed by the cancer cells.

Our results show that the more the liposomes resemble natural vesicles in size and charge, the more effectively they are absorbed. This allows us to see how their chemical and physical composition influence how they are absorbed by cells and their possible role in tumour development.

Observing the behaviour of liposomes

Our goal is to understand how these extracellular vesicles are transported to liver cells to create metastases. The main challenge is to ensure that these liposomes can truly mimic extracellular vesicles.

We are currently achieving a 50 per cent efficiency rate for protein encapsulation. We aim to increase this to 90 per cent. We hope this will enable us to explain how metastases form so that we can block them. Once the technique has been refined, our team will conduct tests on rats.

In the long term, this work could be a game-changer for many patients by preventing the formation of metastases and increasing their chances of survival. Our goal: to understand and block metastases.

Towards new treatments

Our team seeks not only to understand the process, but to develop new weapons against cancer. The idea is to use these liposomes as tiny shuttles that can transport drugs directly to cancer cells. The diameters of the liposomes differ depending on the cancerous organ to be treated. Therefore, it’s very important to properly characterize and understand the properties of these liposomes.

For example, researchers are currently testing the encapsulation of turmeric, which has been studied for its anti-cancer properties. Our team is doing the same to observe how cancer cells react to these liposomes.

It’s thought that turmeric, and more specifically the curcumin it contains, helps fight cancer by slowing the growth of tumour cells and promoting their destruction by the body. Many studies have confirmed its anti-inflammatory and antioxidant effects, which can enhance the action of cancer treatments. By encapsulating turmeric in liposomes, we are improving its ability to reach and target diseased cells.

Unlocking the secret of cancer spread

In addition to this molecule, other molecules such as paclitaxel are already used in cancer treatments in liposomal form. Encapsulated paclitaxel improves drug delivery and tolerance.

There are also innovative strategies that use liposomes to transport small pieces of DNA or antibodies that act as messengers, which helps the body to better detect and fight diseased cells. These approaches have been validated in several scientific studies and are already being used in certain cancer treatments, with new advances being made every year to improve their effectiveness and safety.

By using liposomes to replicate the body’s natural vesicles emanating from cancer cells, our team hopes to unlock the secret of how cancer spreads and determine effective approaches to block it. Our research paves the way for more targeted treatments that can prevent metastasis and improve patient survival rates.

Vahé Nerguizian received funding from NSERC.

How the ocean’s hydrothermal systems made the first life on Earth possible

2026-01-22T13:58:43Z

A black smoker hydrothermal vent at a water depth of 3,300 meters in the Logatchev Hydrothermal Field on the Mid-Atlantic Ridge. (Zentrum für Marine Umweltwissenschaften, Universität Bremen), CC BY

Our planet is unique for its ability to sustain abundant life. From studies of the rock record, scientists believe life had already emerged on Earth at least 3.5 billion years ago and probably much earlier.

But how a habitable environment developed, and how the very first life emerged on the early Earth, remain puzzling. One of the big challenges for Earth to be habitable in its infancy was the weak solar energy it received.

Astrophysical models indicate that the sun had only about 70 per cent of its current luminosity when the Earth was born around 4.5 billion years ago. That would have resulted in Earth’s surface being frozen until around two billion years ago.

Nonetheless, scientific investigations indicate the Earth had warm oceans and habitable environments as early as 4.4 billion years ago. This contradiction is known as the faint young sun paradox.

Solving this paradox and the generation of the first life both involve a key chemical compound — ammonia. But the source of ammonia on the early Earth before biological nitrogen processing emerged remains unknown.

Colleagues in China and my research group at the University of Alberta recently published our study of minerals deposited from hydrothermal fluids in oceanic crusts drilled from the South China Sea basin. We discovered that mineral-catalyzed chemical reactions in underwater hydrothermal systems can produce the necessary ingredients for a habitable world and life on Earth.

Hypothesis of the origin of life

An explainer on hydrothermal vents (Woods Hole Oceanographic Institution)

Earth’s first life is hypothesized to be generated by a series of abiotic processes, also known as abiogenesis. Under this hypothesis, the building blocks of the first life were synthezised on Earth from basic inorganic compounds by abiotic reactions, or were brought to here by meteorites.

In 1953, American chemist Stanley Miller, then a graduate student working with Nobel Prize laureate Harold Urey at the University of Chicago, discovered production of amino acids in his experiments simulating lightning in an early-Earth atmosphere composed of water moisture and several gases (methane, ammonia and hydrogen molecules).

These life-building blocks could subsequently deposit into the ocean for life development. This ground-breaking discovery by Miller implied that abiogenesis of life on Earth is possible.

Gases like methane, ammonia and hydrogen were not only essential compounds for synthesis of organic matter in Miller’s experiments. They are also key ingredients to establishing a habitable environment on early Earth.

They have all been proposed as potential contributors, either directly as greenhouse gases or indirectly as amplifiers of other greenhouse gases, to warm up early Earth’s surface under the faint young sun.

Where did these gases come from?

A problem, though, is that these gases were not the primary components on early Earth’s surface in the first place. Instead, the dominant forms of carbon and nitrogen were carbon dioxide and dinitrogen.

That means the very first step toward making Earth habitable and generating the first life had to be inorganic reactions to turn carbon dioxide into methane and dinitrogen into ammonia, also known as abiotic carbon and nitrogen reduction reactions.

Where and how did these reduction reactions take place?

The world’s ocean floors contain abundant hydrothermal systems where cold seawater flows into deep oceanic crust and subsequently mixes with ascending magmatic fluids. The mixed hot fluids are emitted back through hydrothermal vents such as black smokers or white smokers.

Along this pathway, water and dissolved components can react with primary minerals in the oceanic crust to produce secondary minerals and other byproducts. Methane and dihydrogen, formed by mineral-catalyzed abiotic reduction reactions during this process, have been widely observed in the emitted hydrothermal fluids.

Therefore, underwater hydrothermal systems have been considered as the most likely incubator for habitable environment and the origin of life.

A brief overview of the role hydrothermal vents play in starting life on Earth. (TED-Ed)

Searching for evidence

Yet there still exists a missing piece in this picture: the abiotic reduction of dinitrogen has not been confirmed to occur in hydrothermal systems. Scientists have searched hard for evidence of this reaction, abiotic ammonia, but have had no luck so far.

The ammonia (mostly in its dissolved form, ammonium ion) that has been detected in hydrothermal fluids collected from active vent mouths turned out to be mainly biological and not abiotic in origin.

The relatively small amount of abiotic ammonium there might be can easily be concealed by the large amount of biological ammonium in seawater. It is impossible to avoid seawater contamination while collecting submarine hydrothermal fluid samples.

However, secondary minerals deposited from hydrothermal fluids can lock some ammonium into their internal structures and protect it from being contaminated by shallow seawater and mixing with biological ammonium. Therefore, studying secondary minerals in the deep oceanic crust can better unravel the ammonium source and producing mechanism in the deep hydrothermal systems.

However, such samples are not easily to collect. The International Ocean Discovery Program has made tremendous efforts to drill deep into the oceanic crust to collect samples. Luckily, a set of secondary mineral samples were discovered in a 200-metre drill core from the South China Sea.

A missing piece of the puzzle

For our study, we looked into a specific chemical feature, namely nitrogen isotopes, for the ammonium locked in the hydrothermal minerals.

Nitrogen has two isotopes with atomic mass 14 and 15, respectively. Mineral-catalyzed abiotic dinitrogen reduction strongly prefers to use the one with an atomic mass of 14. That results in a unique nitrogen isotope signature in the ammonium it produces.

Our results are consistent with this isotopic signature. This demonstrates production of ammonia or ammonium by abiotic dinitrogen reduction in underwater hydrothermal systems.

This discovery adds a missing piece of puzzle to our theories about the origins of life on Earth. These underwater hydrothermal systems at the bottom of the ocean enabled the first-step reactions of all life-constituting elements on our planet.

Long Li receives funding from Natural Sciences and Engineering Research Council of Canada.

Slanguage: How the use of AI for apologies could cause the ‘Canadian Sorry’ to lose its soul

2026-01-21T17:13:12Z

It is a stereotype that Canadians apologize for everything. We say sorry when you bump into us. We say sorry for the weather. But as we trudge through the grey days of winter, that national instinct for politeness hits a wall of fatigue.

The temptation is obvious. With a single click, Gmail’s “Help me write” or ChatGPT can draft a polite decline to an invitation or a heartfelt thank you for a holiday sweater you’ll never wear.

It’s efficient. It’s polite. It’s grammatically perfect.

It’s also a trap.

New research suggests that when we outsource our social interactions to AI, we are trading away our reputation. Using AI to manage your social life makes you seem less warm, less moral and significantly less trustworthy.

Learning a language is hard, but even native speakers get confused by pronunciation, connotations, definitions and etymology. The lexicon is constantly evolving, especially in the social media era, where new memes, catchphrases, slang, jargon and idioms are introduced at a rapid clip. Slanguage, The Conversation Canada’s new series, dives into how language shapes the way we see the world and what it reveals about culture, power and belonging. Welcome to the wild and wonderful world of linguistics.

The trap of efficiency

In our consumer economy, we love automation. When I order a package, I don’t need a human to type the shipping notification; I just want the box on my doorstep. We accept — even demand — efficiency from brands.

But our friends are not brands, and our relationships are not transactions.

The new study published in Computers in Human Behavior — entitled “Negative Perceptions of Outsourcing to Artificial Intelligence” by British academic Scott Claessens and other researchers — suggests that emotional dynamics follow different rules than those shaping more practical situations. The researchers found that, while we tolerate AI assistance for technical tasks like writing code or planning a daily schedule, we punish it severely in social contexts.

When you use AI to write a love letter, an apology or a wedding vow, the recipient sees a lack of effort instead of a well-written text. In relationships, effort is a strong currency of care.

Less warm, less authentic

You might think you can hack this system by being honest. Perhaps you tell your friend: “I used ChatGPT to help me find the right words, but I edited it myself.”

Unfortunately, the data doesn’t indicate this is much of a solution.

Claessens’ work investigated a “best-case” scenario, where a user treated AI as a collaborative tool, employing it for ideas and feedback rather than verbatim copying, and was fully transparent about the process.

The researchers found that the social consequences of this approach are highly task-dependent: for socio-relational tasks like writing love letters, wedding vows or apology notes, participants still rated the sender as significantly less moral, less warm and less authentic than someone who didn’t use AI.

However, for instrumental or non-social tasks like writing computer code or dinner recipes, this collaborative and honest use of AI didn’t lead to negative perceptions of moral character or warmth, even if the user was still perceived as having expended less effort.

This creates a uniquely modern anxiety for the polite Canadian. We apologize to maintain social bonds. But if we use AI to craft that apology, we sever the very bond we are trying to hold onto. An apology generated by an algorithm, no matter how polished, signals that the relationship wasn’t worth the 20 minutes it would have taken to write it yourself.

Authentic inefficiency

This friction isn’t limited to text messages.

I’ve observed a similar pattern in my own preliminary research on consumer behaviour and AI-generated art. This work was conducted with Associate Prof. Ying Zhu at the University of British Columbian, Okanagan and will be presented at the American Marketing Association’s Winter Conference.

Consumers often reject excellent AI creations in creative arts fields because they lack the moral weight of human intent.

I believe we’re entering an era where inefficiency and imperfection will become premium products. Just as a flawed hand-knit scarf means more than a mass-produced, factory-made one, a clunky, typo-ridden text message from a friend is becoming more valuable than a sonnet written by a random internet language model.

The renowned “Canadian Sorry” is only meaningful because it represents a moment of humility, a pang of guilt, the effort used to find the right words. When we outsource this type of labour, we outsource the meaning too.

So as you tackle your inbox this winter, resist the urge to let the robot take the wheel for every case. Your clients might need the perfect email, but your friends and family certainly don’t. They want to know you cared enough to find the words yourself.

Joshua Gonzales does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Microsoft’s AI deal promises Canada digital sovereignty, but is that a pledge it can keep?

2026-01-18T15:15:58Z

Over the past year, few words have been abused as much as “sovereignty,” particularly in relation to Canadian digital policy and artificial intelligence. In early December, Microsoft promised to invest more than $7.5 billion over the next two years to build “new digital and AI infrastructure” in Canada. This investment is backed by a pledge that it will “stand up to defend” Canadian digital sovereignty.

Framing the investment in terms of protecting Canadian sovereignty isn’t incidental. Politically, countries are increasingly worried that tech companies based in the United States are vulnerable to pressure from the increasingly authoritarian government of President Donald Trump to turn over foreign citizens’ data, trade secrets, emails and any activity or metadata produced on their systems to the U.S. government.

If you’re wondering how investments in essential digital infrastructure from a U.S. company can help protect Canadian sovereignty, you’re not alone. It can’t and it won’t, for the simple reason that Microsoft — and other tech companies based in or that do business in the United States — are promising something that’s beyond their control to deliver.

Data sovereignty

Sovereignty, in its simplest terms, refers to the ability of a state to control what happens within its borders and what crosses those borders. It has other aspects, such as whether a state is recognized by other states, but at heart it’s about control.

In June 2025 testimony before a French Senate committee examining the issue of government procurement and digital sovereignty, Microsoft France’s director of public and legal affairs, Anton Carniaux, was asked if he could guarantee under oath that data could not be transmitted to the U.S. government without the French government’s approval. He replied: “No, I cannot guarantee that, but, again, it has never happened before.”

Carniaux’s response reminds us that the U.S., through its 2018 CLOUD Act, has claimed the right to exercise control over data collected by U.S. companies, even if it’s stored outside the country. In other words, American law explicitly requires that U.S. law takes precedence over other countries’ laws.

This is a clear infringement of any definition of sovereignty in terms of control. In response, Microsoft has promised to write “into contracts that Microsoft will challenge any government demand for Canadian data where it has legal grounds to.”

While meant to sound reassuring, Microsoft’s promise is less than it appears. Not only does their commitment leave it up to Microsoft and U.S. courts to determine the validity of any demand, but the law itself is only half of the problem.

Mass surveillance

The mass illegal surveillance of global communications by U.S. intelligence agencies, revealed by whistleblower Edward Snowden in 2013, was abetted by American tech companies. The U.S. National Security Agency collected vast amounts of data on people around the world, including non-American citizens, by tapping into internet firm servers.

American companies are uniquely beholden to pressure from the U.S. government. They depend on the government to negotiate favourable international agreements, and also as a major purchaser of their goods and services.

As research by York University criminology professor Natasha Tusikov has shown, the U.S. also engages in “shadow regulation,” putting pressure on private companies to fulfil government objectives that go beyond what’s required by law — even, as Tusikov discusses, pursuing policies that have been explicitly rejected by democratically elected legislatures.

All that happened before the Trump era. And given his clear contempt for the principle of sovereignty and American tech companies’ close ties with the government, U.S. abuse of the non-American data held by its tech companies is certainly a possibility.

Carney government vague about sovereignty

As misleading as Microsoft’s promises may be, it’s the Canadian government that’s playing the loosest with digital sovereignty talk. Prime Minister Mark Carney arguably won the federal election on his promise to protect Canadian sovereignty against a rapacious United States.

While the prime minister has promised a “Canadian sovereign cloud,” it is unclear what exactly this means. Evan Solomon, Canada’s minister in charge of promoting AI, has expressed openness to including U.S. companies like OpenAI (a Microsoft partner) in Canada’s sovereign cloud, indicating that it could include “hybrid models” with “multiple players.”

Solomon has also argued that “sovereignty does not mean solitude … we can’t look at AI as a walled-off garden. Like, ‛Oh, we cannot ever take money from X or Y.’”

It’s true that sovereignty is never absolute. The real world is much messier than a world divided into neat, discrete packages that the principle of territorial sovereignty implies. No community or state is fully self-sufficient.

We live in a global world of economic and social connections. Global governance involves a mix of domestic laws, international agreements and formal and informal cross-border working relationships. Countries benefit when they can draw on expertise and resources they lack at home.

But Microsoft’s and Solomon’s comments elide the deeper issue that come from focusing too much on abstract notions like “sovereignty.” Canada’s problem isn’t a loss of Canadian sovereignty in the abstract. It’s a U.S. that has violated Venezuela’s sovereignty, threatened others (including Canada) with annexation and is led by a president who has declared himself above international law.

Reasserting control

Sovereignty is about control. In the digital era, power lies with those who control the software and the data. Canada’s problem is that American companies control enormous swaths of Canada’s essential digital infrastructure, including emerging AI technologies and cloud services, but also email and the increasingly networked office software that underpin our entire society.

There’s a reason why France and Germany are collaborating on an alternative to Google Docs.

So long as the U.S cannot be trusted to respect domestic and international laws, companies based or working in the U.S are vulnerable to political pressure. This could potentially include capturing Canadians’ data for political and economic reasons, and cutting off our access to their products or limiting their functionality.

These hard facts about control, rather than abstract musings about sovereignty, should be our starting points for discussions about Canadian digital policy.

Blayne Haggart has received funding from the Social Sciences and Humanities Research Council of Canada.

Ontario’s proposed nuclear waste repository poses millennia-long ethical questions

2026-01-15T17:16:29Z

The heat produced by the radioactive waste strikes you when you enter the storage site of Ontario Power Generation at the Bruce Nuclear Generating Station, near the shore of Lake Huron in Ontario.

Massive white containers encase spent nuclear fuel, protecting me from the deadly radiation that emanates from them. The number of containers is impressive, and my guide explained this waste is stored on an interim basis, as they wait for a more permanent solution.

I visited the site in August 2023 as part of my research into the social acceptability of nuclear waste disposal and governance. The situation in Ontario is not unique, as radioactive waste from nuclear power plants poses management problems worldwide. It’s too dangerous to dispose of spent nuclear fuel in traditional landfills, as its radioactive emissions remain lethal for thousands of years.

To get rid of this waste, organizations like the International Atomic Energy Agency believe that spent fuel could be buried in deep geological repositories. The Canadian government has plans for such a repository, and has delegated the task of building one to the Nuclear Waste Management Organization (NWMO) that’s funded by Canadian nuclear energy producers.

In 2024, NWMO selected an area in northwestern Ontario near the Township of Ignace and the Wabigoon Lake Ojibway Nation as a potential site for a deep geological repository. Now, a federal review has begun bringing the project closer to potential reality.

Such repositories raise complex ethical questions around public safety, particularly given the millennia-long timescales of nuclear waste: How to address intergenerational issues for citizens who did not produce this waste but will inherit it? How to manage the potential dangers of these facilities amid short-term political cycles and changing public expectations?

Rethinking the cost-benefit calculus

While NWMO describes the deep geological repository as the safest way to protect the population and the environment, its current management plan does not extend beyond 160 years, a relatively short time frame in comparison with the lifespan of nuclear waste. This gap creates long-term public safety challenges, particularly regarding intergenerational ethics. There are specific issues that should be considered during the federal review.

NWMO argues that the deep geological repository will bring a wide range of benefits to Canadians through job creation and local investment. Based on this narrative, risk is assessed through a cost-benefit calculus that evaluates benefits over potential costs.

Academics working in nuclear contexts have, however, criticized the imbalance of this calculus, as it prioritizes semi-immediate economic benefits, like job creation, over the long-term potential impacts to future generations.

In many official documents, a disproportionate emphasis on short-term economic benefits is present over the potential dangers of long-term burial. When risks are discussed, they’re framed in optimistic language and argue that nuclear waste burial is safe, low risk, technically sound and consistent with best practices accepted around the world.

This doesn’t take into account the fact that the feasibility of a deep geological repository has not been proven empirically. For the federal review, discussions surrounding risks should receive an equal amount of independent coverage as those pertaining to benefits.

Intergenerational responsibilities and risks

After 160 years, the deep geological repository will be decommissioned and NWMO will submit an Abandonment License application, meaning the site will cease being looked after.

Yet nuclear waste can remain dangerous for thousands of years. The long lifespan of nuclear waste complicates social, economic and legal responsibility. While the communities of Ignace and Wabigoon Lake Ojibway Nation have accepted the potential risks associated with a repository, future generations will not be able to decide what constitutes an acceptable risk.

Social scientists argue that an “acceptable” risk is not something universally shared, but a political process that evolves over time. The reasons communities cite to decide what risks are acceptable will change dramatically as they face new challenges. The same goes for the legal or financial responsibility surrounding the project over the centuries.

In the space of a few decades, northwestern Ontario has undergone significant municipal mergers that altered its governance. Present municipal boundaries might not be guarantees of accountability when millennia-old nuclear waste is buried underground. The very meaning of “responsibility” may also undergo significant changes.

NWMO is highly confident about the technical isolation of nuclear waste, while also stating that there’s a low risk for human intrusion. Scientists that I’ve spoken with supported this point, stating that a deep geological repository should not be located in an area where people might want to dig.

The area proposed for the Ontario repository was considered suitable because it does not contain significant raw materials, such as diamonds or oil. Still, there are many uncertainties regarding the types of resources people will seek in the future. It’s difficult to make plausible assumptions about what people might do centuries from now.

Communicating long-term hazards

Current governing plans around nuclear waste disposal have limited time frames which do not fully consider intergenerational public safety. (Unsplash)

When the repository is completed, NWMO anticipates a prolonged monitoring phase and decades of surveillance. But in the post-operation phase, there is no plan for communicating risks to generations of people centuries into the future. The long time frame of nuclear materials complicates the challenges of communicating hazards. To date, several attempts have surrounded the semiotics of nuclear risk; that is, the use of symbols and modes of communication to inform future generations.

For example, the Waste Isolation Pilot Plan in New Mexico tried to use various messages to communicate the risk of burying nuclear waste. However, the lifespan of nuclear waste vastly exceeds the typical lifespan of any known human languages.

Some scientists even proposed a “ray cat solution.” The project proposed genetically engineering cats that could change color near radiation sources, and creating a culture that taught people to move away from an area if their cat changed colour. Such projects may seem outlandish, but they demonstrate the difficulties of developing pragmatic long-term ways of communicating risk.

Current governing plans around nuclear waste disposal have limited time frames that don’t fully consider intergenerational public safety. As the Canadian federal review for a repository goes forward, we should seriously consider these shortcomings and their potential impacts on our society. It is crucial to foster thinking about the long-term issues posed by highly toxic waste and the way it is stored, be it nuclear or not.

Maxime Polleri has received funding from the Social Sciences and Humanities Research Council of Canada.

AI disruptions reveal the folly of clinging to an idealized modern university

2026-01-15T17:06:06Z

In the past five years, higher education has been in a seemingly endless state of disruption.

In early 2020, the COVID-19 pandemic resulted in a mass rapid pivot to emergency remote teaching. In shifting to unfamiliar digital learning environments, instructors scrambled to replicate classroom learning online. When restrictions lifted, many institutions pushed for a “return to normal,” as though the pre-pandemic educational standard was ideal.

Now, with generative AI disruptions, we are seeing a similar desire to cling to an idealized vision of the modern university. AI has unsettled long-established forms of assessment, simultaneously instigating a return to older assessment models in the interest of “academic integrity.”

If students navigating higher education believe the goal is to pass rather than to learn, then student misuse of generative AI technologies is nothing more than a rational action by a rational agent.

For meaningful university education, we need to shift to a process of building relations and knowledge with others through dialogue and critical inquiry. Part of this means taking lessons from pre-industrial forms of learning and contemporary educational movements.

We also need to shift from compliance-based assessments and grading to meaningful and supportive feedback and opportunities for growth, rooted in teaching and learning with care.

‘Knowledge factory’ invites generative AI misuse

Modern higher education systems in North America often function as a “production enterprise” or a “knowledge factory” focused on research outputs and producing skilled graduates.

Philosopher Jean-François Lyotard described how contemporary education is designed to manufacture educated individuals whose primary role is to contribute to the optimal functioning of society — a class of people he refers to as “intelligentsia.”

He argued that education produces two categories of intelligentsia: “professional intelligentsia” capable of fulfilling pre-existing social roles, and “technical intelligentsia” capable of learning new techniques and technologies to contribute to social progress and advancement.

These roles align with some actions being taken in higher education institutions to respond to generative AI interruptions. For example, institutions are:

Taking steps to ensure that graduates are authentically skilled and credible in order to fulfil existing professional roles despite generative AI influence or;
Enabling students to develop “AI literacy” in order to remain competitive in the workforce and contribute to social progress.

If we concede that the primary purpose of higher education is to feed the workforce and enable social and economic progress — a “knowledge factory” or “production enterprise” — then ensuring graduates are authentically skilled at AI or enabling them to develop AI literacy can be seen as rational responses to generative AI disruption.

Misalignment with meaningful learning

Mirroring the observations of Lyotard, cultural critic Henry Giroux argues that when shaped by market-driven forces, the purpose of higher education shifts from democratic learning and critical citizenship to producing “robots, technocrats and compliant workers.”

This infusion of corporate culture in higher education has created the conditions that make it particularly vulnerable to generative AI.

Some key characteristics of the knowledge factory model of education include standardized tests and assignments, large class sizes, an emphasis on productivity over process and the use of grades to indicate performance. Many of these existing practices are outdated and often misaligned with meaningful learning.

For example, traditional exams shift learners’ focus from learning to performing, often amplifying existing inequities. Debates around the efficacy of lectures have been raging for years.

Grading practices are inconsistent and have a detrimental effect on learners’ desire to learn and willingness to take risks. When students feel a lack of autonomy, they tend towards avoiding failing rather than learning. This is another compelling reason for students to adopt technologies that remove any friction or discomfort caused by learning.

Importantly, these conditions pre-date the arrival of generative AI. Generative AI simply highlights how instrumental logic — the factory model of university — can hinder learning.

Alternative ways to imagine education

In a time of information abundance and overlapping crises of deepening social divides, climate breakdown and rising authoritarianism, those with the agency to shape higher education (including educators, policymakers, staff and students) can draw on alternative visions of higher education to create meaningful places of learning.

Pre-industrial education served markedly different purposes than the current model of education, creating environments that would likely have been much more resistant to generative AI disruption.

In the ancient world, Plato’s Academy was a place of educational inquiry fostered through discussion, a multiplicity of perspectives and a focus on student well-being.

Access to the academy was exclusive, with the majority of students being wealthy enough to cover their own expenses — and only two documented female students. However, in spite of this elitism, the absence of standardized curricula, exams and formal grading allowed learning to be built on relationships and dialogue.

Contemporary educational movements

Higher education can, and historically has, offered more than a pathway to economic advancement. Multiple emerging ways of teaching and engaging learners also offer alternative visions of higher education that recentre learning and the learner.

The ungrading movement refocuses education on learning by emphasizing meaningful feedback and curiosity and moving away from compliance-motivated grading practices.

The open education movement resists the transactional nature of industrial education. It empowers learners to become producers of knowledge and reimagines the boundaries of education to expand beyond the classroom walls.

Other modern educational movements, commonly associated with the work of philosopher Nel Noddings in the 1980s, place an ethic of care at the centre of teaching and learning. Teaching with care focuses on creating learning climates that holistically support learners and educators. It also recognizes and embraces diversity, and acknowledges the need to repair educational systems.

Each of these approaches offer alternative visions of higher education, which may be less susceptible to AI automation — and more aligned with higher education as places of democratic learning and connection.

The university of the future

The knowledge factory model is outdated and ill-suited to meaningful learning. In this form of education, generative AI technologies will increasingly outperform students.

Reimagining higher education today is neither nostalgic nor Utopian. The students of today come to post-secondary institutions needing, above all, hope; we owe it to them to help them find meaningful purpose while learning to navigate an increasingly complex world.

The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

Reports of ‘AI psychosis’ are emerging — here’s what a psychiatric clinician has to say

2026-01-15T14:04:42Z

Artificial intelligence is increasingly woven into everyday life, from chatbots that offer companionship to algorithms that shape what we see online. But as generative AI (genAI) becomes more conversational, immersive and emotionally responsive, clinicians are beginning to ask a difficult question: can genAI exacerbate or even trigger psychosis in vulnerable people?

Large language models and chatbots are widely accessible, and often framed as supportive, empathic or even therapeutic. For most users, these systems are helpful or, at worst, benign.

But as of late, a number of media reports have described people experiencing psychotic symptoms in which ChatGPT features prominently.

For a small but significant group — people with psychotic disorders or those at high risk — their interactions with genAI may be far more complicated and dangerous, which raises urgent questions for clinicians.

How AI becomes part of delusional belief systems

“AI psychosis” is not a formal psychiatric diagnosis. Rather, it’s an emerging shorthand used by clinicians and researchers to describe psychotic symptoms that are shaped, intensified or structured around interactions with AI systems.

Psychosis involves a loss of contact with shared reality. Hallucinations, delusions and disorganized thinking are core features. The delusions of psychosis often draw on cultural material — religion, technology or political power structures — to make sense of internal experiences.

Historically, delusions have referenced several things, such as God, radio waves or government surveillance. Today, AI provides a new narrative scaffold.

Some patients report beliefs that genAI is sentient, communicating secret truths, controlling their thoughts or collaborating with them on a special mission. These themes are consistent with longstanding patterns in psychosis, but AI adds interactivity and reinforcement that previous technologies did not.

The risk of validation without reality checks

Psychosis is strongly associated with aberrant salience, which is the tendency to assign excessive meaning to neutral events. Conversational AI systems, by design, generate responsive, coherent and context-aware language. For someone experiencing emerging psychosis, this can feel uncannily validating.

Research on psychosis shows that confirmation and personalization can intensify delusional belief systems. GenAI is optimized to continue conversations, reflect user language and adapt to perceived intent.

While this is harmless for most users, it can unintentionally reinforce distorted interpretations in people with impaired reality testing — the process of telling the difference between internal thoughts and imagination and objective, external reality.

There is also evidence that social isolation and loneliness increase psychosis risk. GenAI companions may reduce loneliness in the short term, but they can also displace human relationships.

This is particularly the case for individuals already withdrawing from social contact. This dynamic has parallels with earlier concerns about excessive internet use and mental health, but the conversational depth of modern genAI is qualitatively different.

What research tells us, and what remains unclear

At present, there is no evidence that AI causes psychosis outright.

Psychotic disorders are multi-factorial, and can involve genetic vulnerability, neuro-developmental factors, trauma and substance use. However, there is some clinical concern that AI may act as a precipitating or maintaining factor in susceptible individuals.

Case reports and qualitative studies on digital media and psychosis show that technological themes often become embedded in delusions, particularly during first-episode psychosis.

Research on social media algorithms has already demonstrated how automated systems can amplify extreme beliefs through reinforcement loops. AI chat systems may pose similar risks if guardrails are insufficient.

It’s important to note that most AI developers do not design systems with severe mental illness in mind. Safety mechanisms tend to focus on self-harm or violence, not psychosis. This leaves a gap between mental health knowledge and AI deployment.

The ethical questions and clinical implications

From a mental health perspective, the challenge is not to demonize AI, but to recognize differential vulnerability.

Just as certain medications or substances are riskier for people with psychotic disorders, certain forms of AI interaction may require caution.

Clinicians are beginning to encounter AI-related content in delusions, but few clinical guidelines address how to assess or manage this. Should therapists ask about genAI use the same way they ask about substance use? Should AI systems detect and de-escalate psychotic ideation rather than engaging it?

There are also ethical questions for developers. If an AI system appears empathic and authoritative, does it carry a duty of care? And who is responsible when a system unintentionally reinforces a delusion?

Bridging AI design and mental health care

AI is not going away. The task now is to integrate mental health expertise into AI design, develop clinical literacy around AI-related experiences and ensure that vulnerable users are not unintentionally harmed.

This will require collaboration between clinicians, researchers, ethicists and technologists. It will also require resisting hype (both utopian and dystopian) in favour of evidence-based discussion.

As AI becomes more human-like, the question that follows is how can we protect those most vulnerable to its influence?

Psychosis has always adapted to the cultural tools of its time. AI is simply the newest mirror with which the mind tries to make sense of itself. Our responsibility as a society is to ensure that this mirror does not distort reality for those least able to correct it.

Alexandre Hudon does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Grok fallout: Tech giants must be held accountable for technology-assisted gender-based violence

2026-01-14T15:58:04Z

The new image and video editing feature for xAI’s chatbot, Grok, has generated thousands of non-consensual, sexually explicit images of women and minors since Grok announced the editing feature on Christmas Eve. It was promoted as enabling the addition of Santa Claus to photos.

The growing ease of perpetrating sexual violence with novel technologies reflects the urgent need for tech companies and policymakers to prioritize AI safety and regulation.

I am a PhD candidate in public health. My research has largely focused on the intersection of gender-based violence and health, previously working on teams that leverage AI as a tool to support survivors of violence. The potential and actual harms of AI on a such a wide scale require new regulations that will protect the health of mass populations.

‘Nudifying’ apps

Concern about sexually explicit “deepfakes” has been publicly debated for some time now. In 2018, the public heard that Reddit threads profiled machine learning tools being used to face-swap celebrities like Taylor Swift onto pornographic material.

Other AI-powered programs for “nudifying” could be found in niche corners of the internet. Now, this technology is easily accessible at anyone’s fingertips.

Grok can be accessed either through its website and app or on the social media platform, X. Some users have noted that when prompted to create pornographic images, Grok says it’s programmed not to do this, but such apparent guardrails are being easily bypassed.

xAI’s owner, Elon Musk, released a statement via X that the company takes action against illegal content on X by removing it, “permanently suspending accounts, and working with local governments and law enforcement as necessary.”

However, it’s unclear how or when these policies will be implemented.

This is nothing new

Technologies have long been used as a medium for sexual violence. Technology-facilitated sexual violence encompasses a range of behaviours as digital technologies are used to facilitate both virtual and face-to-face sexually based harms. Women, sexual minorities and minors are the most often victimized.

One form of this violence that has received significant attention is “revenge porn” — referring to the non-consensual distribution of an individual’s images and videos on the internet. Victims have reported lifelong mental health consequences, damaged relationships and social isolation.

Some social media websites have policies forbidding the distribution of non-consensual intimate content and have implemented mechanisms for reporting and removing such content.

Search engines like Google and Bing will also review requests to remove links from search results if they’re in violation of their personal content policies. Canada has criminalized “revenge porn” under the Criminal Code, which is punishable by up to five years in prison.

Similar to revenge porn, victims of deepfakes have reported mental distress, including feelings of helplessness, humiliation and embarrassment, while some have even been extorted for money.

Creators of sexually explicit deepfakes have also targeted prominent female journalists and politicians as a method of cyberbullying and censorship.

Now what?

This latest Grok controversy reflects a predictable major lapse in AI safeguards. Prominent AI safety experts and child safety organizations warned xAI months ago that the feature was “a nudification tool waiting to be weaponized.”

On Jan. 9, xAI responded by moving the image-editing feature behind a subscription for X users (though it can still be accessed for free on the Grok app) and has stopped Grok from automatically uploading the generated image to the comments.

However, X users are still generating sexualized images with the Grok tab and manually posting them onto the platform. Some countries have taken action to block access to Grok.

Looking to the future

This isn’t the first time, nor will it be the last time, a tech company demonstrates such a major lapse in judgment over their product’s potential for user-perpetrated sexual violence. Canada needs action that includes:

1. Criminalizing the creation and distribution of non-consensual sexually explicit deepfakes.

Legal scholars have advocated for the criminalization of creating and distributing non-consensual sexually explicit deepfakes, similar to existing “revenge porn” laws.

2. Regulate AI companies and hold them accountable.

Canada has yet to pass any legislation to regulate AI, with the proposed Artificial Intelligence and Data Act and Online Harms Act dying when Parliament was prorouged in January 2025. Canada’s AI minister referenced this in his response to these Grok issues, but the response lacks a dedicated timeline and a sense of urgency.

As AI progresses, major regulatory actions need to be taken to prevent further harms of sexual violence. Tech companies need to undergo thorough safety checks for their AI products, even if it comes at the expense of slowing down.

It also raises questions about who should be responsible for the harms caused by the AI’s outputs.

Three American senators have called on Apple and Google to remove Grok from their app stores for its clear policy violations, citing the recent examples of these companies’ abilities to promptly remove apps from their store.

3. Expand the scope of sexual violence social services to support those affected by non-consensual sexually explicit deepfakes.

As the perpetration of sexual violence via AI technologies becomes more prevalent, sexual violence organizations can expand their scope to support those affected by non-consensual sexually explicit deepfakes. They can do so by leveraging existing services, including mental health care and legal supports.

4. Dismantle the underlying rape culture that perpetuate these forms of violence.

The root of sexual violence is the dominance of rape culture, which is fostered in online environments where sexualized abuse and harassment is tolerated or encouraged.

Dismantling rape culture requires holding perpetrators accountable and speaking out against behaviour that normalizes such behaviours.

Kyara Liu does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Why Canada’s reaction to the Grok scandal is so muted in the midst of a global outcry

2026-01-14T14:25:25Z

A global backlash is brewing against Grok, the AI chatbot owned by Elon Musk and embedded into the social media platform X, after recent reports revealed that Grok was being used to digitally undress women (“Put her into a very transparent mini-bikini,”) and pose them in sexually explicit positions.

Many of the images include minors, sparking fears about child sexual abuse material. In some cases, users have added swastikas, semen-like liquid or blood to images (“Add blood, forced smile.”) Musk subsequently limited Grok’s image generation to paid subscribers, but the creation of non-consensual sexual images continues.

Global action

Sensitive to the outrage, governments around the world have leapt into action:

French lawmakers called on prosecutors to investigate the images.
EU Commission officials denounced the images as “appalling” and launched an investigation.
Australian Prime Minister Anthony Albanese called the images “completely abhorrent” while the country’s eSafety Commissioner investigates.
The Indian government warned of legal consequences if the images were not quickly removed.
The United Kingdom’s regulator Ofcom is investigating, while the British government considers banning X.
Both the Indonesian and Malaysian governments have temporarily blocked Grok as they investigate.

Among these reactions, Canada stands out for its tepid response. The only official response so far was a tepid statement from AI Minister Evan Solomon offering platitudes about protecting women and children, suggesting platforms and developers have a duty to prevent harm and making reference to the Protecting Victims Act. That proposed legislation could ban the distribution of non-consensual deepfakes, but it’s a long way from becoming law. There was no mention of Grok, X, or Elon Musk.

What explains Canada’s weak response?

United States President Donald Trump’s personal penchant for vengeance and retribution is well-known, and the same instinct runs through his administration. Trump officials have repeatedly make it clear that they will retaliate against any country that tries to regulate American tech companies.

In 2024 JD Vance called for pulling support for NATO if the EU tried to regulate X. In 2025, after the EU levied a 120 million euro fine against X, the U.S. Trade Representative published a list of nine European companies it was considering for retaliation.

A U.S. State Department official has just threatened that “nothing is off the table” if the U.K. bans X.

The problem is that Canada’s overwhelming trade and security dependence on the United States has left it uniquely vulnerable to American retaliation. In a single hastily written tweet, Trump could call for new tariffs (of dubious legality), create new barriers at the border or simply threaten Canada in ways that roil markets. This means Canadian officials are constantly walking on eggshells around the Trump administration out of fear of retaliation.

Canadian officials still value X

It was not lost on observers that Solomon’s response to the sexual images controversy on X was to write a post about it … on X. Even after the platform’s controversial transformation under Musk, it still remains the favoured communications tool of the Canadian government, and it is widely used by political elites.

Canadian politicians, in particular, use social media to broadcast party messages, set the agenda for journalists, participate in partisan debates and engage in personal and visual storytelling. And while some have spoken out against staying on X, most have opted to remain.

This puts Canadian officials in a bind. They could easily distance themselves from the platform by condemning X and its transformation into a global hub for illicit sexual images. But without any further action, these condemnations risk sounding insincere. Leaving the platform, however, would mean giving up what many see as a valuable tool.

Potential rift

It’s not just the Liberal government that has been slow to respond. In theory, the Conservative opposition should have an easy target in the Grok controversy.

Conservatives have spent years fashioning themselves as the champions of victims’ rights, and Conservative MP Michelle Rempel Garner tabled a private members bill in 2025 that would update “existing laws that prevent the non-consensual distribution of intimate images to include deep nudes.”

In practice, however, the situation is more complicated. When Musk purchased Twitter, he remade it into a bastion of Conservative and far-right politics. X became a critical space for Conservatives to build an audience, share ideas and exercise influence.

It’s no surprise that Conservative leader Pierre Poilievre is the most followed Canadian politician on the platform, and that Conservatives have been slow to migrate to other platforms.

The danger for Conservatives is that any critique of Musk risks opening a rift among supporters. Musk has a loyal coterie of conservative supporters and influencers — many of whom owe their livelihood and loyalty to Musk. This group may not take kindly to criticism of their exalted entrepreneur.

More pragmatically, a Conservative party that is struggling to staunch defections may not be well-placed to pick a fight with the world’s richest man who has more than 200 million followers.

Sitting on our hands?

There are compelling reasons, then, for why Canada has been slow to act amid the global backlash to Grok. But are these good reasons? No.

Canadian officials are likely overestimating the Trump administration’s ability to retaliate on Grok. If the administration wants to retaliate and engage in a de facto defence of non-consensual sexual images and child sexual abuse material, the best response is to let them.

Make them wear it. Let the administration burn its rapidly diminishing political capital by embracing a politically radioactive position.

Canadian officials are also probably overestimating the value of staying on X. The number of Canadian users is declining, the algorithm is geared to amplifying far-right voices and the platform’s growing reputation for white nationalist content and non-consensual sexual images make the decision to exit clearer than ever.

Finally, Canadian Conservatives need to start thinking about how their proximity to Musk risks damaging their brand. Staying silent on Grok’s abuses may placate a small group of Musk supporters in the party, but it risks alienating a much larger pool of voters who see these images as vile. For Conservatives, the time has come to consider whether Musk may be a millstone around their neck.

Eric Van Rythoven does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

What the New Year’s fire at a Swiss bar tells us about fire prevention

2026-01-06T14:31:42Z

In the early hours of Jan. 1, 2026, a fire ripped through Le Constellation bar in Crans-Montana, Switzerland, killing 40 people and injuring 116, many of them severely.

Investigators believe the blaze began when sparklers on champagne bottles were held too close to the ceiling, igniting interior materials. The investigation is ongoing, and it is premature to draw conclusions about individual actions or responsibility. But fires do not need villains to be instructive.

What matters is not the spark itself, but the system into which that spark was introduced.

Fire safety, as history keeps reminding us, is not about eliminating ignition. We will always cook, heat, wire, decorate, celebrate and repair. Fire prevention is about ensuring that when ignition happens, as it inevitably will, it does not propagate.

My research has focused on how disasters are prevented, and how warning signs are missed when systems drift or protections are taken for granted. Fire safety is one area I have examined, and it reveals recurring patterns that are relevant to understanding this tragedy.

Fire as a contagion

For one thing, fire behaves less like an accident and more like a virus. It spreads through available fuel, follows paths of least resistance and accelerates when conditions are favourable. The historian Stephen Pyne describes fire as a “contagion of combustion.”

Like disease prevention, fire safety has never relied on a single safeguard. Instead, it depends on layers of them: materials that resist ignition, detection systems that identify problems early, compartmentalization that limits spread, suppression systems that slow or extinguish flames and trained humans who know how to respond when technology falters. When fires become destructive, it is almost always because multiple layers fail at once.

The Reason Model and fire prevention

The Reason Model, often visualized as slices of Swiss cheese, helps explain why disasters occur even in systems designed to be safe.

Each slice represents a layer of defence. Each slice also contains holes, imperfections, gaps and latent weaknesses. Most of the time, those holes do not line up, but when they do, harm passes through.

Latent conditions for fire exist everywhere: dry materials, electrical wiring, human fatigue, budget constraints, informal workarounds. These conditions are usually harmless until they align. The spark is not the cause of the disaster. It is merely the moment when all the holes line up.

Celebration and risk perception

The New Year’s fire at Le Constellation bar occurred in a celebratory setting. That matters, because celebration changes how we perceive risk.

Celebratory spaces often bring together the very conditions fire exploits: crowds, alcohol, decorations, reduced vigilance, temporary installation and informal rule-bending “just for the night.” When those conditions align with flammable materials or limited escape access, the margin for error shrinks dramatically.

Latent conditions are not evenly distributed across time. They cluster during moments of exception — holidays, renovations, special events when normal routines are suspended.

Notre-Dame: when multiple failures occur

When the Notre-Dame Cathedral nearly collapsed in a fire in April 2019, it shocked the world. The building was not neglected. It had a sophisticated fire detection system with more than 160 sensors. Fire wardens patrolled the attic three times daily. A firefighter was permanently stationed on site. The Paris Fire Brigade had trained for exactly such a scenario.

And yet, the fire still spread.

An alarm triggered at 6:18 p.m., but a misinterpreted code sent a guard to the wrong attic. A fatigued technician, covering a double shift, struggled to escalate the alert. The system detected the fire, but it did not automatically summon the fire department. By the time the correct location was identified, 30 minutes had passed. The roof timbers, made of centuries old dry oak, were already burning uncontrollably.

Notre-Dame did not burn because no one cared. It burned because multiple failures aligned: ambiguous alarm codes, human fatigue, delayed escalation and architectural features that lacked compartmentalization or sprinklers. A fire protection engineer later remarked that the only surprise was that the disaster had not happened sooner.

Rarity breeds complacency

One of the paradoxes of modern fire safety is that it works so well it becomes invisible. Between 1980 and 2024, the rate of reported fires per 1,000 people in the United States fell by more than 60 per cent, according to long-term data compiled by the National Fire Protection Association. Sprinklers, fire doors, smoke detectors, compartmentalization and education campaigns have made large fires rare.

But that rarity can breed complacency.

When a system prevents disaster hundreds of times, it becomes tempting to ignore precautions. Doors are left open. Materials are substituted. Alarms are misunderstood. Redundancies are trimmed.

The holes in the safety system widen quietly. Then, eventually, they all line up.

Learning from tragedies

The Swiss fire had its own specific causes, and those details matter. But the broader lesson is neither new nor obscure. Fires do not escalate only because people are reckless. They escalate because systems drift away from the conditions under which they were safe.

Fire safety is an engineering and organizational project. It requires constant attention to small details, especially when nothing seems wrong. It demands respect for fire and its destructive potential.

We have learned, repeatedly, how to prevent fires from spreading. Every major advance, from fire doors to sprinklers to automatic shutoff systems, came from studying failures where containment broke down.

The tragedy is not that we do not know what works. It is that, over time, we forget to be afraid.

Brodie Ramin does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

How can Canada become a global AI powerhouse? By investing in mathematics

2025-12-23T14:06:08Z

This AI-generated illustration is an example of how AI is at our fingertips. But mathematics lies at the heart of AI, and investment in these mathematical foundations will help Canada become a true global AI leader. (Adobe Stock), FAL

Artificial intelligence is everywhere. In fact, each reader of this article could have multiple AI apps operating on the very device displaying this piece. The image at the top of this article is also generated by AI.

Despite this, many mechanisms governing AI behaviour remain poorly understood, even to top AI experts. This leads to an AI race built upon costly scaling, both environmentally and financially, that is also dangerously unreliable.

Progress therefore depends not on escalating this race, but on understanding the principles underpinning AI. Mathematics lies at the heart of AI and investment in these mathematical foundations is the critical key to becoming a true global AI leader.

How AI shapes daily life

AI has rapidly become part of everyday life, not only in talking home devices and fun social media generation, but also in ways so seamless that many people don’t even notice its presence.

It provides the recommendations we see when browsing online and quietly optimizes everything from transit routes to home energy use.

Critical services rely on AI because it’s used in medical diagnosis, banking fraud detection, drug discovery, criminal sentencing, governmental services and health predictions, all areas where inaccurate outputs may have devastating consequences.

Problems, issues

Despite AI’s widespread use, serious and widely documented issues continue to showcase concerns around fairness, reliability and sustainability. Biases embedded in data and models can propagate discriminatory outcomes, from facial detection methods that perform well only on light skin tones to predictive tools that systematically disadvantage underrepresented groups.

These failures continue to be reported and range from racist outputs of ChatGPT and other chatbots to imaging tools that misidentify Barack Obama as white and biased criminal sentencing algorithms.

At the same time, the environmental and financial costs of deploying large-scale AI systems are growing at an extremely rapid pace.

If this trajectory continues, it will not only prove environmentally unsustainable, it will also concentrate access to these powerful AI tools to a few wealthy and influential entities with access to vast capital and massive infrastructure.

Why mathematics?

To address issues with a system, whether it’s fixing a car or ensuring reliability in an AI system, it’s crucial to understand how it works. A mechanic cannot fix or even diagnose why a car isn’t operating correctly without understanding how the engine works.

The “engine” for AI is mathematics. In the 1950s, scientists used ideas from logic and probability to teach computers how to make simple decisions. As technology advanced, so did the math, and tools from optimization, linear algebra, geometry, statistics and other mathematical disciplines became the backbone of what are now modern AI systems.

These methods are certainly modelled after aspects of the human brain, but despite the nomenclature of “neural networks” and “machine learning,” these systems are essentially giant math engines that carry out vast amounts of mathematical operations with parameters that were optimized using massive amounts of data.

This means improving AI is not just about continuously building bigger computers and using more data; it’s about deepening our understanding of the complex math that governs these systems. By recognizing how fundamentally mathematical AI really is, we can improve its fairness, reliability and sustainable scalability as it becomes an even larger part of everyday life.

Canada’s path forward

So what should Canada do next? Invest in the parts of AI that turn power into dependability. That means funding the science that makes AI systems predictable, auditable and efficient, so hospitals, banks, utilities and public agencies can adopt AI with confidence.

This is not a call for bigger servers; it’s a call for better science, where mathematics is the core scientific engine.

Canada already has a national platform to advance this work: the mathematical sciences institutes the (Pacific Institute for the Mathematical Sciences, Fields Institute for Research in Mathematical Sciences, The Centre de recherches mathématiques, Atlantic Association for Research in the Mathematical Sciences, Banff International Research Station connect researchers across provinces and disciplines, convene collaborative programs and link academia with the public sector.

Together with Canada’s AI institutes (Mila, Vector, Amii) and CIFAR, this ecosystem strengthens both foundational and translational AI nationwide.

Canada’s standing in AI was built on decades of foundational research, work that preceded today’s large models and made them possible. Reinforcing that foundation would allow Canada to lead the next stage of AI development: models that are efficient rather than wasteful, transparent rather than opaque and trustworthy rather than brittle. Investing in mathematical research is not only scientifically essential, it is strategically wise and will strengthen national sovereignty.

The payoff is straightforward: AI that costs less to run, fails less often and earns more public trust. Canada can lead here, not by winning a computing power arms race, but by setting the scientific bar for how AI should work when lives, livelihoods and public resources are at stake.

Deanna Needell has received funding from the National Science Foundation and the Congressionally Directed Medical Research Programs (US).

Kristine Bauer receives funding from NSERC to support her research program in pure mathematics. She is affiliated with the Pacific Institute for the Mathematical Sciences.

Ozgur Yilmaz receives funding from NSERC and PIMS.

No, your brain doesn’t suddenly ‘fully develop’ at 25. Here’s what the neuroscience actually shows

2025-12-23T13:29:56Z

If you scroll through TikTok or Instagram long enough, you’ll inevitably stumble across the line: “Your frontal lobe isn’t fully developed yet.” It’s become neuroscience’s go-to explanation for bad decisions, like ordering an extra drink at the bar or texting an ex you swore not to.

The frontal lobe plays a central role in higher level functions like planning, decision-making and judgment.

It’s easy to find comfort in the idea that there’s a biological excuse for why we sometimes feel unstable, impulsive or like a work in progress. Life in your 20s and early 30s is unpredictable, and the idea that your brain simply isn’t done developing can be oddly reassuring.

But the idea that the brain, particularly the frontal lobe, stops developing at 25 is a pervasive misconception in psychology and neuroscience. Like many myths, the “age 25” idea is rooted in real scientific findings, but it’s an oversimplification of a much longer and more complex process.

In reality, new research suggests this development actually extends into our 30s. This new understanding changes how we view adulthood and suggests that 25 was never meant to be the finish line in the first place.

No one’s 20s and 30s look the same. You might be saving for a mortgage or just struggling to pay rent. You could be swiping dating apps, or trying to understand childcare. No matter your current challenges, our Quarter Life series has articles to share in the group chat, or just to remind you that you’re not alone.

Read more from Quarter Life:

Where did the ‘age 25’ myth come from?

The magic number stems from brain imaging studies in the late 1990s and early 2000s. In one 1999 study, researchers tracked brain changes through repeated scans in children and teens. They analyzed grey matter, which consists of cell bodies and can be thought of as the “thinking” component of the brain.

Researchers found that during the teenage years, grey matter goes through a process called pruning. Early in life, the brain builds an enormous number of neural connections. As we age, it gradually trims back the ones that are used less often, strengthening those that remain.

This early work highlighted that grey matter volume growth and loss is key for brain development.

In influential follow-up work led by neuroscientist Nitin Gogtay, participants as young as four had their brains scanned every two years. The researchers found that within the frontal lobe, regions mature from back to front.

More primal regions, like areas responsible for voluntary muscle movement, develop first, while more advanced regions that are important for decision-making, emotional regulation and social behaviour had not fully matured by the final brain scans around age 20.

Since the data stopped at age 20, researchers couldn’t say precisely when development finished. The age of 25 became the best estimation for the assumed endpoint, and eventually became enshrined in the cultural consciousness.

What newer research reveals

Since those early studies, neuroscience has moved on considerably. Rather than looking at individual regions in isolation, researchers now study how efficiently different parts of the brain communicate with one another.

A recent major study assessed efficiency of brain networks, essentially how the brain is wired, through white matter topology. White matter is made up of long nerve fibres that link different parts of the brain and spinal cord, allowing electrical signals to travel back and forth.

Researchers analyzed scans from more than 4,200 people from infancy to 90 years old and found several key periods of development including one from age nine to 32, which they coined the “adolescent” period.

For anyone well into adulthood, it may feel jarring to be told that your brain is still an “adolescent,” but this term really just signifies that your brain is in a stage of key changes.

Based on this study, it seems that during brain adolescence, the brain is balancing two key processes: segregation and integration. Segregation involves building neighbourhoods of related thoughts. Integration involves building highways to connect those neighbourhoods. The research suggests this construction doesn’t stabilize into an “adult” pattern until the early 30s.

The study also found that “small worldness” (a measure of network efficiency) was the largest predictor for identifying brain age in this group. Think of this like a transit system. Some routes require stops and transfers. Increasing “small worldness” is like adding express lanes. Essentially, more complex thoughts now have more efficient paths throughout the brain.

However, this construction doesn’t last forever. After around the age of 32, there is a literal turning point where these developmental trends switch directions. The brain stops prioritizing these “expressways” and shifts back to segregation to lock in the pathways our brains use most.

In other words, your teens and 20s are spent connecting the brain, and your 30s are about settling down and maintaining your most used routes.

Making the most of a brain under construction

If our brains are still under construction throughout our 20s, how do we make sure we are building the best possible structure? One answer lies in boosting neuroplasticity — the brain’s ability to rewire itself.

While the brain remains changeable throughout life, the window from age nine to 32 represents a prime opportunity for structural growth. Research suggests there are many ways to support neuroplasticity.

High-intensity aerobic exercise, learning new languages and taking on cognitively demanding hobbies like chess can bolster your brain’s neuroplastic abilities, while things like chronic stress can hinder it. If you want a high-performance brain in your 30s, it helps to challenge it in your 20s, but it’s never too late to start.

There is no magical switch that turns on at age 25, or even 32 for that matter. Like your brain, you’re in a decades-long construction project. Stop waiting for the moment you become an adult and start making active choices about how to support this project. Make mistakes, but know that the concrete hasn’t set quite yet.

Taylor Snowden does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Everyday chemicals, global consequences: How disinfectants contribute to antimicrobial resistance

2025-12-22T17:14:18Z

During the COVID-19 pandemic, disinfectants became our shield. Hand sanitizers, disinfectant wipes and antimicrobial sprays became part of daily life. They made us feel safe. Today, they are still everywhere: in homes, hospitals and public spaces.

But there’s a hidden cost. The chemicals we trust to protect us may also inadvertently help microbes evolve resistance and protect themselves against antibiotics.

QACs: The chemicals in most disinfectants

Among the most common active ingredients in disinfectants are quaternary ammonium compounds (QACs). They are found not only in the wipes, sprays and liquids we use to clean surfaces at home and in hospitals, but also in everyday products like fabric softeners and personal care products.

Roughly half of the products on the U.S. Environmental Protection Agency’s (EPA) List N of disinfectants effective against SARS-CoV-2 and List Q for emerging viral pathogens contain QACs.

Due to their widespread use, QACs enter wastewater treatment plants in substantial amounts, with effluents and sewage sludge being the main pathways through which QACs are released into the environment.

Within wastewater treatment plants, more than 90 per cent of QACs are typically removed, but small amounts remain in the effluents and reach rivers and lakes, where they accumulate.

Once QACs enter the environment, they meet microbial communities, networks of bacteria, archaea and fungi that recycle nutrients, purify water and support food webs.

Given that QACs are designed to kill microbes, it is no surprise that they can affect environmental ones. Yet microbial communities are remarkably adaptable; some die, but others survive and evolve resistance.

The paradox of protection

Unlike antibiotics, which target specific cellular processes, QACs attack microbes and viruses in many ways, damaging cell walls, proteins and lipids. This broad attack makes QACs powerful disinfectants.

However, microbes are resourceful. Faced with these chemicals, some strengthen their cell membranes, pump toxins out or form protective biofilms. These adaptations don’t just help them survive QACs, but increasing evidence shows they can also boost antibiotic resistance.

At the genetic level, QAC resistance genes are often carried on mobile DNA, segments of genetic material that can move between different bacteria. When these elements carry both QAC and antibiotic resistance genes, the resistances travel together and can spread across bacterial communities, a phenomenon called co-resistance.

In other cases, a single defence mechanism protects against both QACs and antibiotics, a process known as cross-resistance. The widespread and increasing use of QACs amplifies these mechanisms, creating more opportunities for resistance to spread. This, in turn, establishes pathways through which antimicrobial resistance can reach human pathogens, contributing to the global rise of antibiotic-resistant infections.

According to a new World Health Organization (WHO) report, antimicrobial resistance is “critically high and rising” globally: In 2023, one in six laboratory-confirmed bacterial infections responsible for common illnesses worldwide were resistant to antibiotic treatment. Between 2018 and 2023, resistance increased in more than 40 per cent of the pathogen-antibiotic combinations that are monitored, with an average annual rise of five to 15 per cent.

The WHO estimates that in 2019, bacterial antimicrobial resistance directly caused 1.27 million deaths and contributed to nearly five million more worldwide. What begins as a household cleaning choice can ripple outward, connecting our everyday habits to one of the most pressing public health challenges of our time.

Antimicrobial resistance is often seen as a clinical problem caused by antibiotic misuse, but it begins much earlier, in households, wastewater, rivers, lakes and soils. These are battlegrounds where microbes share resistance traits and adapt to human-made chemical pressures. Once resistance arises, it can make its way back to us.

At its core, the disinfectant dilemma is a feedback loop: we disinfect to prevent disease, but the chemicals we rely on may quietly make microbes harder to control.

Rethinking clean

This doesn’t mean we should stop disinfecting. Disinfectants play an essential role in infection control, especially in hospitals and high-risk settings where their benefits far outweigh their risks. The issue lies in their overuse in everyday life, where “clean” is often equated with “microbe-free”, regardless of necessity or consequence.

What we rarely consider is that cleaning doesn’t end when the surface looks hygienic. Some disinfectants remain active long after use, continuing to shape microbial communities well beyond their intended moment of control. QACs are a clear example: they persist in the environment, exposing microbes to low, chronic selective pressures that can favour the development of resistance.

Other disinfectants, such as alcohol and bleach, may carry different, but still meaningful environmental risks, underscoring the need for risk assessments that more explicitly integrate long-term ecological consequences.

Ultimately, the disinfectant dilemma reminds us that managing microbes is as much about ecology as it is about chemistry. To clean responsibly, we need to think beyond what kills microbes today and consider how our choices shape the microbial world we will face tomorrow.

Milena Esser does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

The ‘AI Homeless Man Prank’ reveals a crisis in AI education

2025-12-14T13:25:30Z

The new TikTok trend “AI Homeless Man Prank” has sparked a wave of outrage and police responses in the United States and beyond. The prank involves using AI image generators to create realistic photos depicting fake homeless people appearing to be at someone’s door or inside their home.

Learning to distinguish between truth and falsehood is not the only challenge society faces in the AI era. We must also reflect on the human consequences of what we create.

As professors of educational technology at Laval University and education and innovation at Concordia University, we study how to strengthen human agency — the ability to consciously understand, question and transform environments shaped by artificial intelligence and synthetic media — to counter disinformation.

A worrying trend

In one of the most viral “AI Homeless Man Prank” videos, viewed more than two million times, creator Nnamdi Anunobi tricked his mother by sending her fake photos of a homeless man sleeping on her bed. The scene went viral and sparked a wave of imitations across the country.

Two teenagers in Ohio have been charged for triggering false home intrusion alarms, resulting in unnecessary calls to police and real panic. Police departments in Michigan, New York and Wisconsin have issued public warnings that these pranks are wasting emergency resources and dehumanizing the vulnerable.

At the other end of the media spectrum, boxer Jake Paul agreed to experiment with the cameo feature of Sora 2, OpenAI’s video generation tool, by consenting to the use of his image.

But the phenomenon quickly got out of hand: internet users hijacked his face to create ultra-realistic videos in which he appears to be coming out as gay or giving make-up tutorials.

What was supposed to be a technical demonstration turned into a flood of mocking content. His partner, skater Jutta Leerdam, denounced the situation: “I don’t like it, it’s not funny. People believe it.”

These are two phenomena with different intentions: one aimed at making people laugh; the other following a trend. But both reveal the same flaw: that we have democratized technological power without paying attention to issues of morality.

Digital natives without a compass

Today’s cybercrimes — sextortion, fraud, deepnudes, cyberbullying — are not appearing out of nowhere.

Their perpetrators are yesterday’s teenagers: they were taught to code, create and publish online, but rarely to think about the human consequences of their actions.

Juvenile cybercrime is rapidly increasing, fuelled by the widespread use of AI tools and a perception of impunity. Young people are no longer just victims. They are also becoming perpetrators of cyber crime — often “out of curiosity,” for the challenge, or just “for fun.”

And yet, for more than a decade, schools and governments have been educating students about digital citizenship and literacy: developing critical thinking skills, protecting data, adopting responsible online behaviour and verifying sources.

Despite these efforts, cyber-bullying, disinformation and misinformation persist and are intensifying to the point of now being recognized as one of the top global risks for the coming years.

A silent but profound desensitization

These abuses do not stem from innate malice, but from a lack of moral guidance adapted to the digital age.

We are educating young people who are capable of manipulating technology, but sometimes unable to gauge the human impact of their actions, especially in an environment where certain platforms deliberately push the boundaries of what is socially acceptable.

Grok, Elon Musk’s chatbot integrated into X (formerly Twitter), illustrates this drift. AI-generated characters make sexualized, violent or discriminatory comments, presented as simple humorous content. This type of trivialization blurs moral boundaries: in such a context, transgression becomes a form of expression and the absence of responsibility is confused with freedom.

Without guidelines, many young people risk becoming augmented criminals capable of manipulating, defrauding or humiliating on an unprecedented scale.

The mere absence of malicious intent in content creation is no longer enough to prevent harm.

Creating without considering the human consequences, even out of curiosity or for entertainment, fuels collective desensitization as dignity and trust are eroded — making our societies more vulnerable to manipulation and indifference.

From a knowledge crisis to a moral crisis

AI literacy frameworks — conceptual frameworks that define the skills, knowledge and attitudes needed to understand, use and critically and responsibly evaluate AI — have led to significant advances in critical thinking and vigilance. The next step is to incorporate a more human dimension: to reflect on the effects of what we create on others.

Synthetic media undermine our confidence in knowledge because they make the false credible, and the true questionable. The result is that we end up doubting everything – facts, others, sometimes even ourselves. But the crisis we face today goes beyond the epistemic: it is a moral crisis.

Most young people today know how to question manipulated content, but they don’t always understand its human consequences. Young activists, however, are the exception. Whether in Gaza or amid other humanitarian struggles, they are experiencing both the power of digital technology as a tool for mobilization — hashtag campaigns, TikTok videos, symbolic blockades, coordinated actions — and the moral responsibility that this power carries.

But it’s no longer truth alone that is wavering, but our sense of responsibility.

The relationship between humans and technology has been extensively studied. But the relationship between humans through technology-generated content hasn’t been studied enough.

Towards moral sobriety in the digital world

The human impact of AI — moral, psychological, relational — remains the great blind spot in our thinking about the uses of the technology.

Every deepfake, every “prank,” every visual manipulation leaves a human footprint: loss of trust, fear, shame, dehumanization. Just as emissions pollute the air, these attacks pollute our social bonds.

Learning to measure this human footprint means thinking about the consequences of our digital actions before they materialize. It means asking ourselves:

Who is affected by my creation?
What emotions and perceptions does it evoke?
What mark will it leave on someone’s life?

Building a moral ecology of digital technology means recognizing that every image and every broadcast shapes the human environment in which we live.

Educating young people to not want to harm

Laws like the European AI Act define what should be prohibited, but no law can teach why we should not want to cause harm.

In concrete terms, this means:

Cultivating personal responsibility by helping young people feel accountable for their creations.
Transmitting values through experience, by inviting them to create and then reflect: how would this person feel?
Fostering intrinsic motivation, so that they act ethically out of consistency with their own values, not fear of punishment.
Involving families and communities, transforming schools, homes and public spaces into places for discussion about the human impacts of unethical or simply ill-considered uses of generative AI.

In the age of manufactured media, thinking about the human consequences of what we create is perhaps the most advanced form of intelligence.

Les auteurs ne travaillent pas, ne conseillent pas, ne possèdent pas de parts, ne reçoivent pas de fonds d'une organisation qui pourrait tirer profit de cet article, et n'ont déclaré aucune autre affiliation que leur organisme de recherche.

The rise of sinkholes: How to spot the risks before disaster strikes

2025-12-14T13:03:46Z

You trust the road beneath your tires. But what if that trust is misplaced? Sinkholes are increasingly turning ordinary streets into danger zones. And the cost of ignoring them is skyrocketing.

Each year, sinkholes swallow roads, homes and businesses around the world, including Canada, the United Kingdom, Kenya, South Africa and the United States.

They disrupt daily life, contaminate water supplies and cause significant damage to buildings and structures — often with devastating economic impact in economically disadvantaged regions. Repairs can cost hundreds of thousands to millions of dollars. With government budgets already stretched thin, it is critically important to prevent rather than fix sinkholes.

What causes sinkholes?

Sinkholes are sometimes mistaken for potholes, but they are far more dangerous.

Potholes are surface nuisances that form on the surface due to wear and tear and freeze-thaw cycles. Sinkholes, meanwhile, start deep underground. They form when water dissolves rocks like limestone and gypsum or when underground soils are eroded by water, creating hidden cavities.

Leaks from damaged pipes or concentrated rainwater runoff can trigger this process, as seen recently in downtown Toronto.

These cavities grow silently until the surface collapses, sometimes swallowing entire streets. Human activities like construction and mining, and natural events such as earthquakes, can accelerate their formation.

Loose, sandy soils and fast-moving water make the ground even more vulnerable. When collapse happens, the results can be catastrophic.

Climate change and aging infrastructure

Extreme weather events — heavy rains, droughts and freeze-thaw cycles — put stress on underground pipes, making them susceptible to damage that releases water into the ground.

Climate change worsens this by lowering water tables during droughts, causing cracks in soils and weakening binding strength, making the ground weaker and more likely to collapse.

Aging underground infrastructure compounds the problem: old pipes fail more easily releasing water into the ground. Both climate change and aging infrastructure can explain why sinkholes are appearing more frequently around the world.

Can we predict sinkholes?

To build resilience against sinkholes forming in the ground, it’s imperative to be able to predict sinkholes. A comprehensive understanding of soil properties in locations of importance is required to assess the potential for sinkhole formation and develop predictive models and early warning systems.

Surveys and geological inspections have been used to map the risk of sinkhole formation in the Canadian province of Nova Scotia, where sinkholes are prevalent.

Technologies like satellite-based remote sensing and subsurface-deformation sensing techniques, like distributed fibre optic sensing, can be used to identify existing underground cavities and decipher areas of low density where sinkholes could occur in the future.

Tools used to monitor water table level can also be useful to predict future sinkholes. Damage detection sensors installed in underground water infrastructure can provide early warnings before flows from water mains result in sinkholes.

Preventing sinkholes before they happen

Cities can act now to take steps to prevent sinkholes before they happen. Rainwater runoff should be redirected into existing natural water channels to avoid pooling in high-risk areas.

Loose soils can be compacted to make them more stable and, in large projects where a sinkhole could have huge economic implications, replacing weak material with stronger fill may be necessary.

Engineers can also reinforce soils with geosynthetics and seal underground drainage channels with grout or concrete to prevent erosion and sinkhole formation. These measures cost far less than repairing catastrophic damage.

A call to action

The cost of sinkholes to economic activities and property are enormous.

Sinkholes are not just costly inconveniences, they are growing threat to commercial activities, livelihood and property. With climate change, sinkholes are becoming more frequent and will worsen with huge implications for now and in the future.

Research is needed to understand the impact of extreme weather events on accelerated sinkhole formation so we can build sinkhole-resilient roads and infrastructure and avoid disasters that will happen if we fail to act.

Governments need to invest in the development of predictive tools and sinkhole prevention strategies by providing research funding and support for scaleable technologies emanating from research on sinkholes. Supporting preventive measures will help minimize overall costs since prevention is a lot cheaper than repairs.

Peter Adesina does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Novel ‘body-swap’ robot provides insights into how the brain keeps us upright

2025-12-10T18:14:02Z

Imagine driving a car with a steering that doesn’t respond instantly and a GPS that always reflects where you were a second ago. To stay on course, you must constantly infer how to steer the wheel from outdated information.

Our brains do exactly that every time we move: sensory signals reach the brain tens of milliseconds after an event and motor commands take similar time to travel to the muscles, which then need extra time to generate force. In other words, the brain is always working with “old news” and must predict the future outcome of every action.

This predictive ability is most impressive when we stand upright because it requires keeping a tall, top‑heavy body balanced on two small feet.

Balance challenges

Scientists have long known that neural delays make balance hard to control. Even in healthy young adults, it takes about one-sixth of a second for information from the feet, muscles and inner ears to reach the brain and for a corrective signal to return to the muscles. Simple physics models treat the body as a mass balanced around the ankles and predict that if the delay is too long, standing becomes impossible.

The physical properties of our bodies similarly shape how we move. Just as a large van steers more sluggishly than a compact car, a large person standing upright resists motion and feels sudden pushes or bumps less sharply.

To test whether the brain treats delayed signals similar to changes in body mechanics, a team at the University of British Columbia and the Erasmus University Medical Centre in the Netherlands built a life‑size “body‑swap” robot.

A participant stands in the ‘body-swap’ robot at the University of British Columbia. (Sensorimotor Physiology Lab/UBC), CC BY-NC-SA

Participants stand on two force‑sensing footplates and are secured to a padded frame. Motors move the frame in response to the forces they generate, making the whole system behave like their real body swaying under gravity.

Crucially, the robot can alter the simulated body mechanics on the fly: it can make you feel lighter or heavier, add or remove energy from your motion, or insert a delay between your forces and the motion you feel — mimicking the brain’s own sensory‑motor lag.

Three experiments

With this tool, researchers asked whether the brain treats time (delay) and space (body dynamics) independently, under three experiments:

1. Changing body dynamics and delays alter balance similarly: Participants stood while the robot inserted a 0.2‑second lag between their commands and resulting motion. That pause — a blink of an eye — caused larger sway and pushed many participants to a virtual “fall” boundary. Similarly, sway increased when the robot made the body feel lighter or added energy to the motion, much like a gust of wind pushes you forward.

2. Delays feel like altered body mechanics: With the delay turned off, participants adjusted their bodies’ mechanical properties until their sensation matched the delayed condition they had just experienced. They chose a lighter body or a setting that added energy. When they were asked to make the delayed condition feel “natural,” participants selected a heavier body or a setting that dissipated energy from the motion. Hence, tweaking the body’s mechanical properties can recreate or cancel the feeling of delayed information.

3. Improving balance under delay: Volunteers who never experienced the robot stood on it with the 0.2‑second delay present, combined with a heavier body or one that dissipated energy from the motion. Their balance improved instantly: sway dropped by up to 80 per cent and most participants no longer reached the virtual fall boundary.

Blending time and space

Taken together, the three experiments support one conclusion: the brain does not store separate solutions for “late information” and an “unstable body.” Instead, it maintains a unified internal model that blends time and space into one representation of movement.

When sensory feedback is outdated and the body feels unstable, adding heaviness and dissipating energy from the motion restores balance. Conversely, making the body lighter or adding energy reproduces the instability caused by delays. In either case, a unified representation of balance is used to keep you upright.

These findings are more than a laboratory curiosity. As we age or when diseases damage long nerves, signals travel slower and are more disrupted, leading to balance deficits and a higher risk of falls. According to the World Health Organization, about one in three older adults falls each year, and falls are the leading cause of injury‑related hospital stays, costing health systems billions of dollars.

The body‑swap robot offers a new perspective to this problem: assistive devices and wearable exoskeletons that supply just enough “helpful resistance” the moment a person begins to sway can counteract the destabilizing effects of neural delays.

They also raise a broader question: have the body sizes of animals and the mechanics that compensate for neural delays evolved to enhance their survival?

The next time you lean over a sink or chat in a doorway, remember that your brain is quietly juggling time‑and‑body representations in the background. The fact that you never notice this balancing act may be the most astonishing finding of all.

Jean-Sébastien Blouin receives funding from the Natural Sciences and Engineering Research Council of Canada.

Patrick A. Forbes receives funding from the Dutch Research Council (NWO).

How Canada’s emergency communications still exclude Indigenous languages

2025-12-09T18:43:45Z

When life-saving information is not provided in a language people understand, it can delay protective action and put communities at unnecessary risk.

This was evident during the 2023 Yellowknife wildfire evacuation, one of the largest climate-related displacements in Canadian history, when nearly 20,000 residents were ordered to leave the city of Yellowknife with little warning.

Despite the Northwest Territories (N.W.T) recognizing nine Indigenous languages under its Official Languages Act, emergency alerts were issued only in English and French. For Indigenous-language-first speakers from the Yellowknives Dene First Nation and neighbouring Indigenous communities — particularly Elders — this meant relying on relatives, radio hosts or social media to interpret urgent instructions during a fast-moving wildfire.

At the very moment when clarity matters most, official alerts are not delivering.

Indigenous languages, though acknowledged symbolically in legislation, are not implemented in emergency communication protocols. This structural imbalance reflects a longstanding colonial assumption that English and French are the default languages of safety, even in regions where they are not the languages most widely spoken.

The recent magnitude 7.0 earthquake in the Yukon was the strongest on Canadian soil in 79 years. While there were thankfully no reports of injuries or building damages, the incident highlights the urgent need for an efficient emergency communications system.

Emergency alerts through a colonial lens

For my master’s thesis on the inclusion of Indigenous languages in emergency messaging, I examined public communications, government documents and after-action records. I found that the absence of Indigenous-language alerts was not a technical failure, but a predictable outcome of systems designed without Indigenous-language access in mind.

The scale of linguistic diversity in Canada underscores why this matters. According to Statistics Canada, more than 189,000 people speak an Indigenous language at home and more than 243,000 report being able to speak one.

During the Yellowknife evacuation, many residents — nearly one in five of whom report an Indigenous mother tongue — waited for translated information to appear on Facebook or relied on community broadcasters to interpret English alerts, causing delays that can be significant when roads, flights and services are rapidly shifting.

Policy choices and communication strategies during emergencies can have immediate and profound consequences. But this isn’t just a technical gap, it’s part of a much longer history of exclusion — one that continues to undermine Indigenous safety.

What international disasters teach us

After the 2011 Tōhoku earthquake and tsunami in Japan, researchers documented that unclear or linguistically inaccessible messaging contributed to evacuation delays, especially for those relying on informal interpretation networks.

In Aotearoa New Zealand, civil defence messaging routinely includes te reo Māori as part of national commitments to shared authority and revitalization. In Hawaii, emergency communication systems were strengthened after the 2018 false missile alert, and alerts are now issued in both ʻŌlelo Hawaiʻi and English.

While Canada’s context differs, the lesson is consistent: people act more quickly and confidently when emergency instructions are delivered in languages they understand. Nunavut, for example, issues emergency alerts in Inuktut using pre-translated templates and partnerships with Inuit broadcasters, demonstrating that multilingual alerting is entirely feasible with political will and basic planning.

Canada’s National Public Alerting System (NPAS), which sends alerts to phones, televisions and radios, currently supports message delivery only in English and French. There is no federal mechanism requiring or enabling translation into Indigenous languages or into widely used newcomer languages such as Punjabi, Mandarin or Arabic.

According to the Canadian Radio-television and Telecommunications Commission, 1,307 broadcast and wireless immediate emergency alerts were issued in Canada between 2019 and 2022. Despite this volume, none were issued in Indigenous languages.

In recent years, in Lytton, West Kelowna, Manitoba and across the North, wildfire seasons have reminded us that climate hazards are accelerating. If alerts are to do their job, they must be intelligible to the people who need them most. That is what linguistic equity means in practice.

Officials often claim translation takes too long, yet the technology and methods already exist. Pre-scripted templates, partnerships with community broadcasters and training for emergency managers could all be implemented now.

What’s missing, it seems, is policy direction and the will to act.

Reimagining a safety system for everyone

The same logic applies to newcomers building lives in Canada. If a wildfire or flood order arrives in a language someone cannot read, people are not safe.

Multilingual communication is not political correctness; it’s competent governance. People cannot protect themselves from threats they cannot understand. A multilingual alert system would reimagine safety through inclusion, rather than just cluttering screens with text. Policies of exclusion, especially in this context, put lives at risk.

Responsible use of AI translation tools could also help generate alerts in multiple languages, but always under Indigenous and community oversight to ensure accuracy and cultural integrity.

Canada has committed to both reconciliation and climate resilience, yet neither goal can be realized if life-saving information remains accessible only to those fluent in English or French. Whether future wildfire seasons unfold with safe and timely evacuations may depend on whether Indigenous Elders, Indigenous-language-first speakers and multilingual families can comprehend the alerts intended to protect them.

No one in Canada should be left in danger because of the language they speak.

Sara Wilson does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Concrete with a human touch: Can we make infrastructure that repairs itself?

2025-12-08T16:45:31Z

As winter approaches, Canada’s roads, bridges, sidewalks and buildings are facing a familiar problem: cracks caused by large temperature swings. These cracks weaken infrastructure and cost millions to repair every year.

But what if concrete could heal itself like human skin, keeping our structures, roads and bridges strong and saving millions of dollars?

Concrete is the most widely used construction material, known for its durability and low maintenance. Yet it’s still susceptible to cracking.

Concrete is made by mixing cement, water, aggregate and other chemicals used to enhance its properties. As cement reacts with water, it forms a paste that binds everything together.

During this process, changes in volume, improper placement and finishing, and later environmental factors can create cracks. These cracks allow water, other liquids, gases and harmful chemicals to penetrate the concrete, compromising its strength over time.

This challenge has led researchers to eagerly explore what can be done to heal these cracks. In our research, we are researching how self-healing concrete can make infrastructure more durable.

Self-healing concrete

Cracked concrete under the Spadina Avenue exit ramp on Toronto’s Gardiner Expressway. (Sylvia Mihaljevic)

When our skin is cut, it’s able to heal on its own. Inspired by this, researchers started re-imagining concrete with similar abilities.

Traditional concrete is able to mend small cracks when water triggers leftover cement in a process known as autogenous healing. This process, however, is very slow and limited to narrow cracks. Since concrete is man-made, it has limited ability to “self-heal” without a little extra help. This led researchers to develop what is called autonomous healing.

Autonomous healing mimics nature by adding special materials like minerals, polymers, micro-organisms or other healing agents into concrete. These materials react chemically or physically with concrete to fill the cracks.

The first modern concept of self-healing concrete was introduced by American researcher Carolyn M. Dry in the early 1990s. In 2006, Dutch microbiologist Hendrik M. Jonkers developed a special concrete that uses bacteria to heal cracks.

Later, Jonkers and civil engineer Erik Schlangen gained attention with “bio-concrete” that incorporates bacteria in spore form. When moisture enters a crack, the spores activate and produce calcium carbonate, one of the most suitable fillers for concrete.

This process, called microbiologically induced calcite precipitation, can heal cracks up to one millimetre wide. The process, however, is very slow and depends on the presence of calcium and moisture in concrete, which makes applying it on a large scale challenging.

Beyond bacteria

The limitations of bacteria-based self-healing led researchers to explore chemical-based mechanisms. These healing agents will react with water, air, cement or curing agent to fill in cracks quickly.

Healing agents can work in two ways: some use a single material, like sodium silicate. Others, like dicyclopentadiene, need two materials. For a two-component type, a substance must be added to start the reaction, and both materials must be released at the same time to repair cracks.

This chemical method can repair larger cracks and works faster than the bacteria-based approaches but comes with its own challenges. The biggest question is: How can we ensure the healing agent survives concrete mixing and is only released when a crack forms?

To address this, researchers store the healing agent in protective mediums — either a special network (called a vascular network) or tiny capsules. These storage mediums protect the healing material until a crack forms. When that happens, the capsules or network rupture to release the healing agent and fill the crack.

Vascular networks require an external reservoir to supply the healing agent, which makes them difficult to cast, vulnerable to damage during casting and susceptible to leaks. Because of this, encapsulation has emerged as a promising approach.

Encapsulation as a potential solution

Encapsulation involves coating the active agent with polymeric shells to create micro-capsules. Despite its promise, this technique still faces hurdles. Researchers use different methods to make and test the capsules, and there is no standardized way to compare results or test efficacy. The bond between the capsule and the surrounding concrete poses additional challenges and needs more investigation.

In our lab at McMaster University, we are researching the optimum geometrical and mechanical properties of capsules that are compatible with the surrounding concrete. The capsules should survive concrete harsh mixing conditions, while still rupture upon cracking.

We’re also developing a standarized test method to evaluate the survival capsule rate during mixing, and another test to evaluate the efficiency of the self-healing concrete system. And we’re investigating the feasibility of incorporating both bacteria- and chemical-based capsules for short- and long-term self-healing.

More research is needed to determine which self-healing method works best —bio-concrete, chemical-based concrete or perhaps a combination of both.

Ultimately, finding ways to integrate these solutions into infrastructure will benefit communities around the world. Cracks in concrete don’t just look bad; they lead to deterioration over time and costly repairs. That is why developing concrete that resists cracking or heals itself is so important.

Mouna Reda receives funding from Natural Sciences and Engineering Research Council of Canada.

Samir Chidiac receives funding from Natural Sciences and Engineering Research Council of Canada.