Imagine two identical spoons. One is hand-wrought from silver by a skilled metalworker. The other, a base-metal facsimile, was mass-produced by a machine. Which would you value more? Most of us would say the handmade spoon.

In 1899, more than a century ago, American economist and sociologist Thorstein Veblen used this very example to explain how we assign value, or his theory of conspicuous consumption, in which he contended that bourgeois consumption was driven primarily by a desire to display wealth to others. Even if these spoons were indistinguishable, explained Veblen, the hand-made spoon, once identified, would be more highly valued.

This is in part because “the hand-wrought spoon gratifies our taste, our sense of the beautiful, while that made by machinery out of base metal has no useful office beyond a brute efficiency.” But for Veblen there is another factor more important than any aesthetic judgment: costliness.

The hand-wrought spoon is preferred above all, Veblen suggested, because it is a means of demonstrating wealth. However, as we enter a world in which almost anything, including art, writing and music, can be machine-wrought, it seems that Veblen may have misjudged his spoons.

We don’t value human creations solely for their beauty or their price tag. We also value them because they embody deliberate labour and expertise.

AI-generated writing is judged differently

Our own research has shown that even highly trained writing educators cannot reliably distinguish between AI-generated and human-written essays. In fact, one study has shown that general audiences may actually prefer blander AI-generated poetry over more difficult, human-written poetry.

But while public taste may favour the simple and formulaic, the disclosure of artificial authorship is enough to make most people recoil.

In a recent study involving a series of experiments, participants were asked to compare pieces of AI-generated creative writing, including poetry and fiction. In each case, they were told that some passages were human-written and some were AI-generated. Across 16 experiments, respondents consistently devalued the writing labelled as AI-generated.

The authors of the study call this the “AI disclosure penalty.” It is possible to conclude from the study that audiences unfairly judge AI-generated content, but we disagree. This bias towards human creation is inherent to our relationship with art. When people believe something was made by a machine, they like it less.

Some argue that AI can democratize creativity by lowering barriers to production and enabling more people to participate in cultural expression. But the evidence suggests that when authorship becomes effortless, perceived value declines.

The importance of effort and experience

Art costs something. Both John Milton and James Joyce believed that their writing had cost them their eyesight. John Keats believed that the emotional exertion of writing poetry would worsen his tuberculosis and cost him his life. They kept writing anyway. We resent the machine because its creations cost it nothing.

When an algorithm generates a story about heartbreak or an essay on human struggle, it is trading in stolen emotions. AI has never felt pain, suffered a loss or wrestled with the frustration of a blank page, so its output, no matter how technically smooth, feels fundamentally deceptive.

People hate the idea of being moved by a parlour trick. In addition, many of us have a deep, instinctive revulsion to the industrialization of our inner lives. As Joanna Maciejewska observed, “I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes.”

We happily accept machines stamping out our car parts and toasters because efficiency is the goal, but applying that same cold logic to human expression strips away the vulnerability, risk and stakes that make art mean anything in the first place.

This becomes more consequential as AI-generated content floods the digital media landscape.

Why human work is becoming more valuable

Our media ecosystem has evolved so that paying directly for much of the content we consume is optional. In an era of streaming music, television and film, we rarely own the product we consume, and creators receive pennies on the dollar compared to previous economic models.

To make matters worse, media companies are increasingly pushing AI-generated content in the form of tens of thousands of social media posts, books, podcasts and videos every day and encouraging artists and content creators to supercharge the quantity of their output by relying on AI.

Much of this output is highly formulaic — produced at scale and designed for rapid, low-engagement consumption. It is an endless, flavourless paste of clichés and nonsense, meant to be mindlessly consumed by doomscrolling thumbs and immediately forgotten. Despite working in an era in which payment is optional amid a deluge of slop, many artists, journalists and writers are making a living because enough of their audience chooses to support the work of real human creators.

The “AI disclosure penalty” reminds us that the consumption of art is not tied to purely aesthetic considerations but involves a need to connect with and appreciate the effort and labour of others.

Consumers have long been willing to pay more for goods labelled “handmade,” “handcrafted,” “artisanal” or “bespoke” on the understanding that those goods were made using traditional techniques that took more effort and human skill.

As generative AI turns writing, art and digital media into frictionless, infinitely replicable outputs, human cognitive effort is undergoing a profound shift. It is becoming an artisanal good that consumers must choose to support and value.

The Industrial Revolution transformed hand-made furniture and hand-woven textiles into premium markers of craftsmanship and authenticity. The AI revolution is doing something similar for intellectual and creative labour — audiences are beginning to place a premium not necessarily on the competent execution of a poem or an essay, which a machine can generate in seconds, but on the invisible friction, the lived experience and the deliberate toil of the human mind behind it.

In a landscape increasingly saturated with instant content, the verified effort of a human creator is shifting from a baseline expectation to a highly coveted, bespoke quality. Ultimately, what we value about art is not whether it’s perfect, but its ability to connect us with another human being.

Nathan Murray has received funding for his research from the Social Sciences and Humanities Research Council of Canada (SSHRC).

Elisa Tersigni has received funding from the Social Sciences and Humanities Research Council of Canada (SSHRC).

When the world shut down due to the COVID-19 pandemic in early 2020, another crisis quietly grew behind closed doors. Reports from around the globe suggested that violence against women and girls was increasing. Governments, nongovernmental organizations and advocates began referring to the phenomenon as a “shadow pandemic.”

To determine whether these headlines and informal reports reflected reality, we led the first-ever comprehensive review of studies tracking violence against women and girls during infectious disease outbreaks across low- and middle-income countries. We focused on those regions because less research on the topic has been done there, and women and girls face specific risks, such as child marriage, that are less prevalent in wealthier nations.

Our findings, published in BMJ Global Health and co-authored with UNICEF, are both clear and concerning: Violence against women and girls tends to increase during outbreaks, and the very measures used to control disease spread can lead to that rise.

Across 53 studies measuring changes in violence against women and girls in the first year of the COVID-19 pandemic, the majority found increases, with some studies showing no change and relatively few showing declines. This pattern held across different types of violence – for example, physical domestic violence, sexual domestic violence, psychological violence or online violence – particularly committed within the home.

But even more striking was how little evidence there was from other infectious disease outbreaks. Despite the growing frequency of global health emergencies, only one study examined violence against women and girls during an outbreak other than COVID-19, specifically examining violence in Sierra Leone during both Ebola and COVID-19.

How outbreaks contribute to gender-based violence

Infectious disease outbreaks do more than spread illness. They can disrupt economies, burden health systems and reshape daily life. These shifts can amplify existing inequalities and, in many cases, increase the risk of violence.

Our research identified five key pathways through which outbreaks contribute to violence against women and girls.

First, job loss, reduced income and financial stress were the most consistently identified contributors to violence. When households experience economic strain, tensions rise – and women and girls often bear the consequences. In some contexts, economic stress was linked not only to intimate partner violence but also to harmful practices like child marriage.

Second, movement restrictions like lockdowns and quarantines can trap women and girls with abusive partners or family members. While these outbreak response measures are designed to reduce disease transmission, they can also isolate women from social networks and limit opportunities to seek both formal and informal help.

Third, deeming certain services as nonessential reduces people’s access to support. During COVID-19, many health, social and legal services were scaled back or became harder to access. School closures also meant that girls in some contexts faced increased risks of exploitation, early pregnancy or forced marriage.

Fourth, perpetrators may use women’s and girls’ fear of infection to control or manipulate them. For example, men sometimes discouraged their partners from leaving the home or seeking care in order to avoid disease risk.

Finally, women’s and girls’ past experiences with health systems can influence their intention to seek services in the future. In settings affected by earlier outbreaks, such as the 2014 Ebola outbreak, mistrust of health services discouraged some survivors from seeking care after experiencing violence, especially if doing so might lead to quarantine or mistreatment.

These pathways are not isolated. They often interact and reinforce one another, creating conditions in which violence becomes more likely during crises.

Building better evidence

Public health emergencies are becoming more frequent, and measures like lockdowns and limiting access to schools, clinics and other services can have unintended consequences. Our findings show that protecting women and girls needs to be part of how public health experts respond to outbreaks from the start and not something to address only after violence has already increased.

Tracking the issue in different types of outbreaks – such as cholera, influenza or Ebola – could help determine which policy responses are most protective.

But even within COVID-19 research, we uncovered important limitations. First, most studies focused on adult women, with far less attention to girls. And second, many studies relied on metrics such as the number of hotline calls or clinic visits, which can be misleading. A drop in reports does not necessarily mean a drop in violence; it may reflect reduced access to services or greater barriers to reporting.

Despite the data gaps we uncovered, our study already points to targeted strategies that can protect women and girls: reducing households’ financial stress, making services safe and easy to reach, ensuring girls’ continued access to school, and building stronger community support.

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

Voters in Wales will soon go to the polls to elect members of an expanded Senedd (Welsh parliament) under a new proportional voting system. As the campaign has developed, public service broadcasters have sought not only to report events but to educate, inform and engage audiences with an unfamiliar electoral process.

Our analysis suggests they are increasingly doing so through digital platforms. We analysed all election news content produced online and on social media by major broadcasters between April 8 and April 24, including BBC Wales, ITV Wales, S4C, Channel 4 and Sky News.

The findings point to a move towards formats designed for audiences who are more likely to encounter news online than through traditional television.

This matters because people increasingly come across political content passively, through algorithmically curated feeds rather than actively seeking it out. In that environment, the type of content produced – and how it’s presented – can play a decisive role in shaping public understanding of the election.

One prominent feature of digital coverage has been the use of explainers. These aim to demystify the election by breaking down how the Senedd works, how the voting system has changed and which policy areas are devolved to Wales or reserved to Westminster.

Many of these explainers adopt a more informal and accessible tone than their broadcast equivalents. They’re designed to cut through in fast-moving social media feeds where political information competes for attention.

A significant proportion focus on policy. Of the 19 explainers identified in our analysis, seven centred on specific issues, most commonly immigration. This reflects persistent public confusion about where responsibility lies.

Read more: Voters in Wales face Senedd election amid confusion over who holds power over what

Our recent survey found that nearly a third of people in Wales did not know immigration is controlled by the UK government. Against that backdrop, broadcasters have often made this distinction explicit. In 82% of online and social media items mentioning immigration, journalists clearly stated that responsibility lies with Westminster.

Broadcasters have also used explainers to clarify changes to the electoral system. This includes the move to a closed-list proportional system. Public awareness of this change remains low, however. Only 7% of respondents in our survey correctly identified the system, while 58% said they did not know.

Meet the leaders

Alongside explainers, broadcasters have used digital formats to introduce audiences to the leaders of Wales’s six main political parties. This has reinforced the campaign’s increasingly presidential tone, with party leaders dominating media appearances.

In a devolved context, this is not always straightforward, given the presence of both UK-wide and Welsh political figures. But digital formats have provided new ways to foreground Welsh leaders.

Short, one-to-one interviews have become an important feature. Formats such as the BBC’s Quickfire Questions and ITV’s Chippy Chats mix light-touch prompts – like “What song have you got on repeat?” – with more substantive questions about policy priorities.

These formats inject personality into political coverage. Leaders are presented not only as decision-makers but as people with interests and personalities. This is particularly significant given relatively low public awareness of Welsh political figures.

Our recent survey found that fewer than half of respondents could identify the leader of Plaid Cymru, Rhun ap Iorwerth, despite the fact he could become the next first minister.

At the same time, the informal tone has not entirely displaced scrutiny. In ITV’s Chippy Chats for example, the Welsh Liberal Democrat leader Jane Dodds was challenged on her voting record in the Senedd. It’s a reminder that accountability can still be built into more conversational formats.

Informing voters in a digital campaign

Taken together, these approaches suggest broadcasters are using digital platforms in distinct and complementary ways. Explainers aim to address gaps in public knowledge. One-to-one interviews make political leaders more visible and relatable.

This reflects a broader transformation in how election coverage is produced and consumed. As more people encounter political information online, public service broadcasters play an increasingly important role in countering misinformation and improving understanding of politics and public affairs.

The challenge is now to strike the right balance. Broadcasters must produce content that engages audiences. But they shouldn’t lose sight of the need to inform them and to scrutinise the claims made by political parties.

Keighley Perkins receives funding from AHRC for research into broadcasters' impartiality.

Maxwell Modell receives funding from the AHRC for research into broadcasters' impartiality.

Stephen Cushion has received funding from the BBC Trust, Ofcom, AHRC, BA, ESRC and Welsh Government.

Eighty-one years after Adolf Hitler died by his own hand in a Berlin bunker, a viral video on TikTok shows an AI-generated vision of the Nazi dictator standing in Antarctica, shoulders broad and face smiling, sipping a White Monster Energy drink while Men at Work’s iconic song Down Under plays.

It’s an absurd image, but one that makes sense in the context of the “Agartha” trend on TikTok, which is quietly bringing white supremacist narratives into the mainstream to be seen by millions of users.

The modern myth of Agartha, a supposed utopia hidden inside the hollow Earth, was constructed from older pieces by esoteric authors after the second world war. It blends “Aryan” white supremacist themes with ideas of an occult SS and Third Reich spaceships.

Literal belief in hollow-Earth myths or Nazi UFOs is not the point. Instead, it’s an aesthetic – one that can host both coded far-right messages and explicit ones, fused with pop-cultural references such as the White Monster Energy meme.

Mainstreaming through borderline content

Agartha videos on social media are “awful but lawful”: the content is objectionable but legal. It allows extremists to embed their narratives into mainstream social media spaces, without triggering moderation or outright rejection by the audience. As a result, they can reach large, young audiences.

To understand how the underlying esoteric myths are used, we analysed a network of more than 43,000 Agartha-related TikTok videos and closely studied selected examples. This analysis is part of an ongoing project led by researchers at Neu-Ulm University of Applied Sciences in Germany. The goal is to understand how extremists abuse platform features to carry radical narratives into the mainstream.

We identified four key mechanisms which far-right actors use to push radical narratives to unsuspecting audiences: aesthetic camouflage, dog-whistles and split-second glimpses, network building, and weaponised irony. Let’s unpack each of these terms.

Aesthetic camouflage

Moderation systems on social media platforms aim to remove overt extremist propaganda. These systems work imperfectly, but overt propaganda is unlikely to reach mass audiences before being removed.

Instead, far-right actors often use generative AI to mask racial ideology behind seemingly benign tropes from science fiction and fantasy. This allows a “de-demonising” of their ideas. Elf-like depictions of the “Aryan” inhabitants of Agartha, or footage of an underground utopia, make the idea of a white ethnostate seem palatable.

The engaging aesthetic keeps people watching longer. In turn, this triggers the algorithm to push the video to wider audiences.

Dog whistles and split-second provocations

Far-right actors infuse their content with dog whistles that communicate to a certain audience without triggering moderation.

Our research identified recurring visual symbols in Agartha videos. “Raw milk” signals white supremacy. The number 271, which appears in hundreds of the videos, is a code for Holocaust denial.

Agartha videos also sometimes include overt extremist markers such as the Hakenkreuz (the Nazi symbol of the “hooked cross” or swastika). They are often flashed for only fractions of a second.

This is a calculated provocation. It tests and pushes platform boundaries, normalising the presence of far-right markers and slowly desensitising viewers. At the same time, successful inclusion of forbidden symbols in videos with viral reach serves as a badge of honour within the in-group.

Building network bridges

The Agartha community is not an isolated echo chamber. Our analysis shows the Agartha network connects to others on TikTok. Around 87% of these connections come via the inclusion of mundane, mainstream hashtags on Agartha videos.

Extremists may deliberately “hijack” benign, high-traffic hashtags, such as #roblox or #loganpaul, to push their content into everyday feeds. Agartha is also strongly connected to gym and fitness content. Hashtags such as #gymtok serve as a bridge, potentially funnelling users towards radical narratives.

Here, we also observed targeted appeals. Agartha videos would commonly include hashtags associated with “looksmaxxing” – a trend for extreme physical self-optimisation – to push their videos towards audiences seen as potentially susceptible, such as insecure young men. Well-known looksmaxxers were also sometimes depicted in Agartha videos.

Weaponised irony

As the example of Adolf Hitler drinking a Monster Energy illustrates, Agartha videos rely heavily on absurd situations, frequently co-opting mainstream social media trends.

Because White Monster Energy is a popular meme within non-extremist spaces, users familiar with the trend are particularly likely to get algorithmic recommendations fused with extremist narratives.

Similarly, actors superimpose mainstream figures into Agartha to force association, such as YouTuber Logan Paul or actor Mads Mikkelsen.

Such co-opting is also done through music, with many hollow-Earth edits set to Men at Work’s 1980s hit Down Under. The track serves as a darkly ironic nod to the subterranean utopia.

This increases the chances for algorithmic amplification while providing creators with plausible deniability. Criticisms can be dismissed as a misunderstanding of dark humour or current trends.

Recognising the threat

Agartha is more than just a digital resurfacing of fringe occultism. It is a blueprint for how to design extremist content for algorithmically curated short-video platforms.

It’s also a reminder that extremist content on social media does not exist in isolation. Instead, it lives in what researchers call “hybridised spaces”, where users move in and out of extremist discourse. In such spaces, borderline content, outright extremism, mundane trends and humour blend seamlessly – and participants may find their mainstream interests lead them to radical narratives.

Marten Risius receives funding through the Distinguished Professorship Program via the Bavarian Hightech Agenda from the Bavarian Ministry of Sciences and Arts.

Christopher David 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.

Canada’s health systems began shifting from paper charts to electronic health records decades ago. These records hold patients’ critical health information, including medications, diagnoses, clinical notes, test results, specialist consults and plans for care.

Our research, published today in the Canadian Medical Association Journal, raises major concerns about the state of these electronic health records nationwide.

In most provinces and territories, information is currently siloed in separate software programs in different offices, designed by multiple vendors with differing standards. This fragments patients’ health records across services and leaves clinicians without the information they need to provide safe care.

This is harming patients, costing taxpayers $9.4 billion annually and hindering health-system improvement.

Canada’s missed opportunity

Ideally, patients’ health information should follow them over time and across locations. Some might assume that’s how it works now. After all, hotel chains remember whether we prefer foam or feather pillows, no matter what country we are in. Uber ratings follow us everywhere.

Unfortunately, in health care, things aren’t so seamless. In the rush to abandon paper charts and transition to electronic records, Canada missed a major opportunity for standardization.

Without an overarching plan, clinics, hospitals and jurisdictions chose from dozens of incompatible platforms sold by vendors competing for market share, without considering the need for personal health information to follow the patient.

A provincial and territorial legislative focus on the privacy of patient records has also fostered an environment that splinters patient information between health services.

The Connected Care Scorecard

Collecting, tracking and exchanging patients’ health information is key to safe, co-ordinated care. In some jurisdictions, like Taiwan, electronic health records from different vendors dock securely together. If a family doctor changes a medication, then pharmacy, hospital and specialist records are automatically updated. A treatment plan from a specialist lands directly in a family doctor’s electronic record, without need for faxing, scanning or uploading.

In Canada, hospitals, specialists and primary-care services still rely heavily on fax machines and mail, rather than automated, instant, accurate data exchange.

As part of our research, we created a Connected Care Scorecard that reveals where each province and territory stands in connecting its health records.

In British Columbia, for example, dozens of incompatible electronic health record systems are used in community clinics alone. Hospitals, even within the same health authority, run on different platforms. A patient who visits an emergency room in downtown Vancouver will have to tell their story again if they later seek care in Burnaby. Clinicians may end up retesting for illnesses already ruled out.

Prince Edward Island does much better — with one electronic health record uniting all hospitals and a single platform for primary-care clinics. The hospital record feeds information into primary care so details are available for follow-up.

Interoperability matters

Connected, integrated electronic health records allow all clinicians to work together on a common plan. Sharing patient information is critical for team-based care. It improves outcomes like medication safety and enables patients’ access to records, making them part of the care team.

Most jurisdictions do have patient portals where some people can see portions of their health information, like lab results or prescriptions. However, a 2025 study found that only 13.2 per cent of adult Canadians have electronic access to such records.

Despite tremendous hype and opportunity to improve care through artificial intelligence, most health systems can’t use it at scale. That’s largely because the opportunities it offers — assisting with diagnoses and prompting clinicians to order the tests and treatments patients need — are wholly dependent on ready access to comprehensive, accurate patient health data.

Interoperable electronic health records would also help health systems access population-based information to inform planning. Data could help predict disease outbreaks and spot bottlenecks in hospital flow. It could improve cancer care and ensure patients with the greatest needs are prioritized.

Our research shows that although most jurisdictions use some hospital data for planning, information in electronic health records, especially from primary care, rarely gets used to improve health systems. This has long-term implications: you can’t manage what you don’t measure.

All of this adds up to massive costs for taxpayers, patients and clinicians.

Common health data standards

The federal government recently reintroduced the proposed Connected Care for Canadians Act, which would require vendors to adopt common standards for exchanging information across systems. It’s a solid first step, but more is needed.

Most importantly, governments must establish clear accountability — nationally, provincially and territorially — for health data oversight. This must balance minimizing privacy breaches with limiting all other forms of harm arising from disconnected records, including damage to patients, clinicians and health systems.

Jurisdictions must also establish common health data standards, tools and incentives to improve data coordination.

Our challenge is not adopting electronic health records, but connecting them. Without that, our investment simply won’t pay off. Care will continue to suffer.

Dr. Ewan Affleck, physician, senior medical advisor in health informatics at the College of Physicians and Surgeons of Alberta and chair of Networked Health, co-authored this article.

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.

The first Monday in May marks the annual Met Gala: a collision of celebrities, designers and cultural icons. Established in 1948, the gala was originally a high-society event held to raise money for the Costume Institute of The Metropolitan Museum of Art.

When former editor-in-chief of Vogue Anna Wintour took over in 1995, she shifted the focus from New York’s elites to celebrities, launching it into a fashion juggernaut.

Each year brings a new theme and new dress code. The theme reflects the Costume Institute’s latest exhibition (which opens the following day). The dress code translates this theme into creative direction for gala attendees.

This year’s theme is Costume Art, and the dress code is Fashion is Art. These ideas showcase fashion as an embodied art form, and explore the historical connection between clothing, the body, the wearer and art.

So, is fashion art? And if so, at what point do clothes transform from something practical to something artistic?

Is fashion art?

Throughout his career, German fashion designer Karl Lagerfeld (1933–2019) upheld the separation of fashion and art. “Art is art, fashion is fashion”, he said.

Lagerfeld’s words were based on a distinction that is commonly understood in the art world between fine art and decorative art.

Fine art is a creative expression designed to elicit an emotional or intellectual response. Artists can work on a single piece for years to create something unique. Traditionally, this category has included paintings, sculpture and poetry.

Decorative art is aesthetically pleasing, but also functional, commercial and mass produced. Examples include home decoration and fashion.

Unlike fine artists, decorative artists or designers generally don’t have the luxury of time, and must continually produce products for market consumption. For these reasons, Lagerfeld didn’t see fashion as art.

Conversely, pop artist Andy Warhol (1928–87) declared: “fashion is more art than art is”.

Warhol’s works were defined by themes of pop culture, consumerism, capitalism and the mass media. They held a mirror to society. Fashion does this too. In addition to being emotional, intellectual and creative, it can reveal the norms and values of a society.

Warhol’s art often crossed into the fashion world through collaborations with designers such as Diane Von Furstenberg and Halston.

Italian fashion designer Elsa Schiaparelli (1890–1973) also saw the merit of fashion as art, stating “designing is not a profession but an art”.

Schiaparelli was one of the earliest designers to challenge the distinction between art and fashion. Her works are currently on display at London’s Victoria and Albert Museum, as part of a broader trend of museums and galleries showcasing haute couture as art in its own right.

Haute couture (which translates to “high dressmaking”) is exclusive, high-end fashion that is different from mass-produced ready-to-wear clothing.

One of the first major haute couture exhibits came in 2011 from the Met itself. Over three months, more than 600,000 people visited Alexander McQueen: Savage Beauty, making it one of the Met’s most visited exhibits in history.

Public appetite has only grown since then. Last year, the Louvre Couture exhibit in Paris received more than one million visitors.

Read more: How self-taught, self-made mavericks Vivienne Westwood and Rei Kawakubo redefined punk

Fashion and modern society

Haute couture may be art, but what about everyday fashion? Can that be art too?

Designer John Galliano (1960–) suggested as much when he said, “the joy of dressing is an art”.

Dressing is an active practice and is vital for participation in society – not just for the sake of modesty, but because attire speaks of identity. Clothing designates how people want to be perceived, and can be an important marker of gender, social status, political affiliation and heritage.

Haute couture artists are also becoming more accessible to the public, reflecting a societal shift that recognises – and even craves – fashion as art.

John Galliano was the lead designer at Christian Dior from 1997 to 2011, the so-called “golden age of haute couture”. He is currently partnered with fast-fashion giant Zara in a two-year collaboration deal.

Perhaps then, fashion becomes art when it transcends functionality and becomes performative, creative or inspirational.

Interpreting Met Gala fashions

So how might we approach judging fashion as art at this year’s Met Gala?

First, ask yourself if the outfit evokes emotion. Not just awe or joy – but even shock, hate or fascination. The primary purpose of art is to elicit feeling.

In 2022, Kim Kardashian sparked outrage when she wore Marilyn Monroe’s famous “Happy Birthday, Mr President” dress to the gala.

The theme that year was In America: An Anthology of Fashion. For many people, Monroe and her famous gown represented the height of American culture.

Kim’s use of the dress sparked broader conversations about historical objects, ethics and celebrity culture. Some also accused her of damaging it.

As you watch this year’s gala, it’s worth examining whether any of the outfits stimulate a thought or conversation about politics, history, technology or culture.

Designers often use colours, textiles and shapes to express something about society. These messages may be subtle, or at times quite explicit.

In 2021, American politician Alexandria Ocasio-Cortez wore a white gown with “tax the rich” written on the back, referencing the extreme wealth disparity in the United States.

Fashion reflects who we are, and the world we live in. If that isn’t art, I don’t know what is.

Grace Waye-Harris 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.

People in the UK are now spending fewer years in good health than they did a decade ago, according to a new analysis by the Health Foundation. The UK now sits near the bottom of a 21-country comparison, ahead only of the US.

A drop in healthy life expectancy is explained through many causes: obesity, alcohol, drugs, suicide, chronic disease, poverty and widening inequality. But one of the most powerful causes sits atop them all: housing. Where and how people live is one of the main factors explaining how health risks are created and distributed across society.

The UK Housing Review is an annual independent review of housing policy and evidence, written by housing experts and published by the Chartered Institute of Housing. Its latest edition, which we contributed to, identifies several interrelated ways that housing affects health.

A key one is affordability – housing costs shape where people can live, whether they can heat their homes, whether they can afford food and transport, whether they can move for work, whether they can leave unsafe or unsuitable housing and whether they live with chronic financial stress.

In the UK, housing costs are high by historical standards and poor housing remains widespread. The review notes that private rents are now at their highest recorded share of earnings, while millions of homes in England still contain serious health and safety hazards.

When housing is unaffordable, people are forced to make tradeoffs. For example, trading affordability for damp or overcrowded homes. They cut back on heating, food, medication, transport and social participation. They move further from public services, work and support networks. Affordability problems also force many people into cheaper, less secure, tenancies.

Poor housing quality directly shapes health. Cold, damp, mould, disrepair, poor ventilation and unsafe homes are directly linked to respiratory illness, cardiovascular risk, mental health problems and reduced wellbeing.

Read more: Cold homes increase the risk of severe mental health problems – new study

The Building Research Establishment, an independent research organisation, has estimated that poor housing costs the NHS in England £1.4 billion each year. More than half of this is attributed to cold homes, which increase the risk of respiratory illness, cardiovascular problems and poor mental health. They are especially dangerous for older people, babies and people with existing health conditions.

But the wider costs are even greater. Poor sleep, stress, disrupted schooling, insecure work, social isolation and caring strain all affect mental and physical health. They increase pressure on families and, over time, on health, education and social care systems.

Historically in the UK, social housing has provided some protection to people unable to access good quality affordable housing in the open market. But the stock of social rented housing in the UK has declined. This means that people are increasingly dependent on (often expensive) market rental, where the quality, size and location of housing depend much more directly on income.

The rise of the private rented sector this century has meant that more households are exposed, not just to higher housing costs, but also to shorter tenancies and fewer protections than social housing traditionally provided.

The Renters’ Rights Act increases security, but does not remove “no fault” evictions altogether and does little to protect tenants from economic pressures that can result in eviction. The cognitive burden of worrying about eviction, arrears, repairs or the next rent increase is a direct health risk.

Recent evidence also suggests that insecure housing can result in measurably faster biological ageing, equivalent to the effects of more traditional health concerns like smoking.

Additional weeks of biological ageing per year from different factors

The number of people living in temporary accommodation has risen dramatically, reaching over 130,000 households at the beginning of 2025. This is a 156% increase compared with 2010, largely driven by the poor affordability and insecurity of the private rented sector and lack of social housing. Temporary accommodation is inadequate housing, particularly for children. Living in temporary accommodation was a contributing factor in the deaths of at least 104 children in England between 2019 and 2025, 76 of whom were under one year of age.

This is not about housing quality alone. Temporary accommodation reflects multiple risks brought together: poverty, overcrowding, poor conditions, instability, lack of space for safe infant sleep, poor access to services and wider racial and social inequality. The National Child Mortality Database identifies temporary accommodation as a contributing factor to vulnerability, ill health or death, not necessarily as the sole cause. Emerging evidence also links temporary accommodation with stillbirth and neonatal death.

Read more: Insecure renting ages you faster than owning a home, unemployment or obesity. Better housing policy can change this

Housing health inequality

ONS data shows a very large difference in healthy life expectancy between the most and least deprived areas. In 2022-24, healthy life expectancy in the most deprived areas of England was just 49.8 years for men and 48.2 years for women, compared with 69.2 and 68.5 years in the least deprived areas.

Housing contributes to this difference, determining whether people live in homes that support recovery or deepen stress, whether children grow up in stable and safe environments, and whether older people can remain warm and independent.

If the government is serious about its stated aim to “halve the gap in healthy life expectancy between the richest and poorest regions”, housing policy must become health policy.

That means investing in social housing, enforcing decent standards in the private rented sector, making homes warmer, safer and more accessible, and recognising temporary accommodation, overcrowding and insecurity as public health failures, not just housing management problems.

It also means changing the way that success is measured. Housing policy is too often judged by supply numbers, prices or tenure outcomes. These matter, but they are incomplete. A healthy housing system should also be judged by whether people can live in homes that are affordable, secure, decent, suitable and resilient to climate change.

The decline in healthy life expectancy is a warning light. It tells us that the UK is not only failing to keep people well for longer, it is failing to provide the foundations of health.

Emma Baker receives funding from the Economic and Social Research Council, the Australian Research Council, The National Health and Medical Research Council, and the Australian Housing and Urban Research Institute.

Amy Clair receives funding from the Australian Research Council and the Australian Housing and Urban Research Institute.

Mark Stephens receives funding from ESRC, the EU/Innovate UK and the Australian Housing and Urban Research Institute (AHURI).

A glance at recent front pages of many British newspapers leaves no doubt about the stridency of their views on net zero.

On January 13, for instance, the Express said the government must “Tell truth on ‘fantasy’ cost of net zero”, while the Mail’s headline on the same day used the same idea of “fantasy figures”. A few weeks later, a Telegraph headline claimed “Labour’s net zero extremism is ripping the heart out of Britain”.

But how representative are these headlines of wider coverage? To find out, colleagues and I analysed nearly 500 articles published over four months in 2023 across nine UK newspapers (both right- and left-leaning), looking at pieces where net zero appeared in the headline.

We focused on the presence of statements which were factually inaccurate, or misleading (defined as the omission of a credible counter-argument).

Outright inaccuracies were relatively rare. We found 22 examples, partly because we used a narrow definition. But misleading claims were very common.

This was especially true in opinion and editorial pieces. In four right-wing outlets – the Telegraph, Mail, Express and Sun – more than 70% of such articles contained at least one misleading statement.

Because a single misleading statement may not be representative of an overall article – perhaps appearing in a quote – we then looked at those articles where there was a pattern, containing at least three misleading statements.

We found 50 such articles, of which 92% were published in the right-wing press, and the vast majority in editorials and opinion pieces. Of the editorials and opinion pieces we flagged at the Telegraph, Mail, Express and Sun, between 39% and 60% included at least three misleading statements.

Articles which contain at least three misleading statements:

The most common misleading statements concerned the potentially high cost of net zero, the various ways the policy was being implemented, and claims about the unfair distribution of costs. These claims were often presented without acknowledging opposing evidence or arguments – for example, that the costs of inaction were also high or possibly higher, or that experts dispute the figures presented in the article.

By contrast, left-wing publications were more likely to mention the high costs of inaction and the potential co-benefits of net zero such as improved health or better air quality.

In this context, remember that in July 2025 the UK government’s Office for Budget Responsibility found that the cost of bringing emissions down to net zero is significantly lower than the economic damages of failing to act. It also found those net zero costs will be much lower than previously expected.

Scrutiny – but fairer and better-informed

This isn’t a call for newspapers and journalists to avoid scrutinising net zero. It’s a policy that will be funded in part by British taxpayers, and may impose significant and uneven costs on different sectors of the population.

But coverage that focuses only on these costs in isolation, or that cherry picks data to support a single view, risks giving readers an incomplete picture. Fairer and better-informed coverage would mention on a regular basis the in-depth findings of a range of experts on the costs of inaction and the co-benefits of action.

The Times, for example, shows that it is possible to quote experts from two sides. In our 2023 sample we found several articles, including some in right-leaning newspapers, where the high cost of net zero is mentioned alongside the benefits of taking action, or that also added the qualification that many climate experts dispute the high costs.

A final thought: in its March 2026 report, the UK’s official advisory Climate Change Committee said that the “cost” of cutting UK emissions to net zero could be less than the cost of a single fossil-fuel price shock, while a net-zero economy would be almost completely protected from future spikes.

I looked in vain for a front-page headline in the Sun, Express or Mail screaming that reaching net zero would be cheaper for the UK than a fossil fuel crisis, such as the one triggered by the war on Iran.

James Painter receives funding from the Grantham Research Institute on climate change and the environment, London School of Economics.

For children with autism spectrum disorder and with an intellectual disability, the options for improving communication and social skills are limited.

Talking therapies and behavioural programmes can help some children develop these skills, but they depend on specialists who are in short supply – even in wealthy countries.

Around 30-35% of autistic children have an intellectual disability, according to research from the US. They are less likely to get treatment than those without one (in part because doctors lack confidence managing their needs and insurance coverage for intellectual disability is patchy) despite having greater needs and placing heavier demands on their families. It is a group that researchers often overlook.

That gap motivated us to test a different kind of intervention: using brief, targeted magnetic pulses to stimulate specific parts of the brain. The technique, known as non-invasive brain stimulation or neuromodulation, involves no surgery, no anaesthetic and no drugs.

A device held close to the scalp generates a rapidly changing magnetic field that passes harmlessly through the skull and stimulates the activity of neurons underneath. It has been used for years to treat depression, and researchers have increasingly been exploring whether it might also help with the social and communication difficulties that are a key symptom of autism.

The version we tested uses a technique called theta-burst stimulation, which delivers pulses in rapid clusters rather than one at a time. This makes each session much shorter than conventional approaches, which is a significant practical advantage when you are asking young children to sit still and cooperate.

In our study, published in the BMJ, each session lasted only a few minutes, and the full course ran over just five days. One group of children received real stimulation, another received a sham version. In the sham treatment, the equipment was applied in the same way and delivered vibrations, but no active pulses were delivered. That way, we could compare results without either group knowing what they’d received, which helps keep the findings reliable.

One hundred and ninety-four children took part, with an average age of around six and a half years. Roughly half had IQ scores below 70, which is typically described as the low-functioning range, though all scored above 50 – the minimum needed to ensure a reliable diagnosis and meaningful participation in the study.

Parents filled in a questionnaire about their child’s social communication, before the treatment, right after, and again a month later.

The improvements seen after five days were still there after a month, and the size of the effect was large by the standards of clinical research. Children also showed gains in language ability.

No serious side-effects were reported and all minor side-effects resolved without treatment.

Early days

Children were recruited from multiple sites by advertisements posted in outpatients clinics and through local clinical registries. All legal guardians gave written consent.

Children with intellectual disability are so often left out of trials of this kind that the evidence for treating them has remained seriously lacking. That this trial included them at all – and in significant numbers – is itself noteworthy. But it is only a first step.

It is still unclear how long the benefits last beyond a month, how many sessions would be needed to maintain them, or how the approach would work when moved from a research setting into an ordinary clinic.

Brain stimulation is not a replacement for behavioural support, and the equipment needed is not cheap or universally available. But conventional approaches – where they exist at all – often require daily sessions over several weeks with a professional, which carries its own costs in time, money and specialist input.

A five-day course is a different proposition. For families who are already stretched, even modest and durable gains in a child’s ability to communicate could matter enormously to them and their families and greatly improve their wellbeing and quality of life.

Barbara Jacquelyn Sahakian receives funding from the Wellcome Trust. Her research work is conducted within the NIHR Cambridge Biomedical Research Centre (BRC) Mental Health and Neurodegeneration Themes. She receives Royalties from Cambridge University Press for Brain Boost: Healthy Habits for a Happier Life.

Christelle Langley is funded by the Wellcome Trust. Her research work is conducted within the NIHR Cambridge Biomedical Research Centre (BRC) Mental Health and Neurodegeneration Themes. She receives Royalties from Cambridge University Press for Brain Boost: Healthy Habits for a Happier Life.

Fei Li receives funding from the National Natural Science Foundation of China. She is affiliated with Department of Developmental and behavioral pediatrics, Society of Pediatrics, Chinese Medical Association.

Qiang Luo 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.

Artificial intelligence is transforming our world and financial services are no exception. AI is reshaping the personal banking sector but where does it currently stand on gender parity, transparency and fairness?

When someone applies for a loan today, there is a growing chance that no human ever reads their application. A data-driven algorithm decides whether they qualify, how much they can borrow, and how risky they are considered, often in a matter of seconds and without explanation, quietly shaping financial opportunities in ways most people never see but feel in their everyday lives.

These systems are usually presented as neutral tools: faster than people, more consistent, less prone to prejudice.

In a sector long criticised for opacity and bias, that promise is appealing and frequently echoed in industry and policy debates. But that promise rests on a fragile assumption, rarely made explicit, that the data these systems learn from reflects everyone’s lives equally.

A recent report by the EU Agency for Fundamental Rights, based on fieldwork in five member states, examined how high-risk AI systems are governed under the EU AI Act in areas such as employment, public benefits and law enforcement. It found a striking gap between legal ambition and practice: while risks of discrimination are broadly acknowledged, providers and deployers often lack the tools, expertise and guidance to assess them systematically. Self-assessments tend to be inconsistent, and oversight remains thin.

This is an important issue. When the data feeding these systems fails to capture the reality of women’s financial lives with the same depth and accuracy as men’s, the result is not just a technical shortcoming but a structural distortion, one that shapes who gets access to credit, on what terms, and with what long-term consequences. For AI-driven finance to be fair, women must first be “visible” in the data on which these systems rely.

Algorithms do not judge fairness or ask whether an outcome makes sense, but estimate what is most likely to be correct based on the data they are given, drawing patterns and projecting them forward. When data is incomplete or distorted, the system’s conclusions rest on shaky assumptions from the start.

If women are underrepresented, poorly measured, or never analysed separately from men, the system cannot see unequal outcomes, and what it cannot see, it cannot correct. Bias is simply carried forward and made routine.

This dynamic is easy to miss when discussions stay at the level of models and regulation, but its effects become clear as soon as automated systems are observed in practice. Across different countries, evidence shows how quickly inequality can be embedded in algorithmic decisions, not because systems are designed to discriminate, but because they faithfully reproduce the distortions already present in the data they learn from.

Kenya offers a telling illustration. According to published studies, a widely used digital lending algorithm consistently offered women smaller loans than men, in some cases by more than a third, despite stronger repayment performance. The system did not single women out deliberately: it simply learned from data shaped by long-standing social and economic disparities, and then applied those patterns at scale.

What matters in this example is not Kenya itself, but what the case makes visible. The algorithm did exactly what it was designed to do, learning from past behaviour and applying those patterns consistently, yet without the ability to distinguish between women’s and men’s outcomes, there was no way to detect that inequality was being reproduced in real time. The problem was not automation, but blindness.

How can finance overcome the gender blind spot?

That is where sex-disaggregated data becomes essential. By sorting financial data by gender, regulators, financial institutions, and technology designers can uncover the impacts of automated systems, identify who has access to finance, and pinpoint areas where outcomes begin to diverge. Without that visibility, gender gaps remain hidden, and hidden gaps have a habit of becoming permanent. In digital finance, data is “a girl’s best friend”, not as a slogan, but as a practical condition for accountability.

Most financial institutions already record a customer’s gender as part of basic identification. On paper, the information is there, embedded in routine reporting and basic customer records. In practice, however, recording a variable is not the same as using it. In many countries, the sex of the customer appears in databases but is never analysed, reported, or monitored by supervisors, including in core supervisory frameworks such as prudential reporting. Too often, the data already exists, but it is collected, filed away, and then quietly ignored. The problem lies not in what can be done, but in what is done.

Fairer finance: developing countries are leading the way

The picture looks very different in countries often assumed to have fewer resources. In parts of Latin America and Africa, regulators have required sex-disaggregated reporting for years and regularly publish data on gender gaps in finance.

In Chile, financial authorities have tracked gender differences in loans and deposits for more than two decades, publishing regular sex-disaggregated financial statistics.

In Mexico, regulators combine bank data with national household surveys to understand how women and men use financial services and how they perform as borrowers.

That visibility has had practical consequences. In Mexico, supervisory data showed that women’s loans were smaller but less risky, evidence that fed into changes in loan loss provisioning rules.

In Chile, the data revealed that equal access to accounts did not translate into equal outcomes in savings or insurance, prompting more targeted policy responses. Once these gaps became visible, they became far harder to ignore.

Seen from this perspective, the situation in many high-income economies looks less like a technical lag and more like an institutional hesitation. In much of Europe, gender data remains voluntary or fragmented despite advanced data infrastructures, a failure not of technical capacity but of institutional choice. My upcoming policy paper “Data Are a Girl’s Best Friends: Tackling Digital Financial Inequality Through Sex‑Disaggregated Data”, due to be published in May explores this.

As artificial intelligence becomes more deeply embedded in financial decision-making, that choice becomes harder to defend. At a time when Europe is implementing the EU AI Act and debating how to regulate algorithmic decision-making in finance, the absence of systematic gender data raises a basic question: how can fairness be monitored if the data needed to detect inequality is never analysed?

Making women visible in the data is not symbolic. Without it, fair finance is little more than a claim.

A weekly e-mail in English featuring expertise from scholars and researchers. It provides an introduction to the diversity of research coming out of the continent and considers some of the key issues facing European countries. Get the newsletter!

Eliana Canavesio est membre de Volt Europa.

Anusuyabai Pandekar and her daughter-in-law Mandabai sit facing each other beside a stone grindmill. The mill is still. No grain rests between its stones. No flour gathers at the edges. Instead it sits between them like an object from another time.

One of the women begins to sing. The other joins. The melody carries the rhythm of a labour no longer being done, cyclical and without clear beginning or end:

It is raining heavily, let the soil become wet.

Women go to the fields, carrying baskets of bhakri (bread).

The pre-monsoon rain is beating down on the fields.

Under the jasmine tree, the ploughman is working with the drill-plough.

Scenes like the one this song describes, once common across rural western India, now belong increasingly to the archive. Hand-grinding has given way to electric mills. The work that once informed these songs has thinned out, leaving behind recordings, fragments and memory.

Accounts of drought and environmental change rarely include such voices. In official records and news reports, what is measured often overshadows what is lived. Climate change is typically explained through numbers, including emissions targets, temperature thresholds and rainfall variability. This data is necessary. But it cannot capture how change is inhabited: how it settles into bodies, reshapes routines and presses into everyday life.

Long before climate science named the crisis, women were registering these shifts in another language – song.

Climate, labour and everyday life

Across the world, women’s work songs function as informal archives of environmental change. Emerging from repetitive labour – including grinding, pounding, planting and carrying – they register shifts in seasons, resources and survival long before these enter formal records.

I began to understand this during my doctoral work in 2020 and 2021. I was researching labour arrangements within the sugar industry in drought-affected regions of western India. Policy reports described rainfall deficits, groundwater depletion and crop loss. But women spoke instead of work – walking further for water, delaying planting and stretching food across uncertain seasons.

Their voices extended beyond conversation into an unexpected archive – The Grindmill Songs Project. First documented in the 1990s and now hosted by the People’s Archive of Rural India, the project brings together around 100,000 songs organised by people, places and themes. I used this archive alongside ethnographic interviews to trace labour, marriage and drought in the sugarcane industry, where women’s voices were largely absent from official records.

Here, labour and environmental strain were articulated with a precision often absent from formal accounts. Climate was not abstract; it was embedded in the rhythms of work.

The climate crisis has a communications problem. How do we tell stories that move people – not just to fear the future, but to imagine and build a better one? This article is part of Climate Storytelling, a series exploring how arts and science can join forces to spark understanding, hope and action.

The water-guzzling sugarcane crop, around which the region’s economy turns, surfaced repeatedly in both speech and song. It appeared as a metaphor for happiness, for domestic violence, even for dowry; a substance moving between fields and households, binding labour, desire and coercion. Environmental stress did not stand apart from these concerns, but moved through them. As one song goes:

A daughter’s existence is like a sack of sugar

Father got his daughter married, he became a merchant

Another describes married life through the language of extraction:

Father says, daughter, how are you treated by your in-laws

Like a 12-year-old sugarcane crushed in the sugar-mill

A broader pattern emerges from this context. Across regions, environmental change is first encountered through its effects on labour, and only later abstracted into data. Comparable dynamics appear elsewhere. In west African farming communities, songs synchronise collective labour while expressing shared experience of seasonal uncertainty. In Malawi, during famine, women sang:

Koke kolole … pull, pull hard, pull the clouds –

why does the rain not come?

Our dead fathers, what have we done?

Forgive us … do you want us to die?

Send us rain.

Here, ecological crisis is framed as a breakdown within a moral and social order. Such songs interpret environmental failure through relationships between the living and the dead and between obligation and neglect.

On the Swahili coast, fishing songs similarly accompany sailing and net-making, embedding weather knowledge, labour discipline and social commentary within everyday maritime life. These songs accompany work, but they also organise it, giving rhythm to collective effort while encoding knowledge about seasons, risk and survival.

This relationship between labour and environment extends across very different histories. In the Caribbean, work songs bear the imprint of plantation economies shaped by extraction and environmental vulnerability. In Latin America, women’s traditions carry histories of colonial labour within their rhythms.

In Colombia’s San Basilio de Palenque, women still sing as they coax peanuts from rain-softened soil, gathering food, language and memory in the same gesture. Elsewhere, songs track movement itself: young men leaving with the dry-season wind, rivers in flood separating families.

Along cold North Sea coasts, herring workers, known as the “gutters”, sang Gaelic work songs in the 19th century while gutting fish at speed, their rhythms coordinating labour under harsh conditions. Beyond work, women also composed laments that dwelt on separation from men at sea.

Listening to climate differently

These songs describe hardship. But they also make it perceptible, situating environmental stress within labour, social relations and obligation. Climate change follows existing inequalities. In many contexts, its earliest effects are absorbed through women’s work, through longer hours, shifting responsibilities and increased strain.

Importantly, these songs were not intentionally composed as records of environmental change. They emerge from labour, relationships and survival. Yet because women’s work is so closely tied to land, water and season, environmental shifts are registered within them, often indirectly, as part of their lived experience.

Work songs therefore offer a distinct kind of record. Against archives that have historically privileged elite and male voices, they preserve forms of knowledge grounded in everyday labour.

But the conditions that sustained such singing are fading. Mechanisation and the decline of collective work have reduced the spaces in which these songs were produced and shared, with many now confined to ritual settings such as weddings and childbirth gatherings. As these practices decline, so too do the forms of knowledge embedded within them.

Listening to these songs does not replace data-driven, scientific knowledge about climate change. It complements it by making visible dimensions of change that are otherwise difficult to capture, including the reorganisation of labour, the strain on relationships and the uncertainty of survival.

Reetika Revathy Subramanian received funding from the Gates Cambridge Trust for her doctoral research at the University of Cambridge Centre for Gender Studies (2019–2024), on which this article primarily draws.

There was a time when Elon Musk and Sam Altman were friends. But the two tech billionaires are now embroiled in a bitter legal battle in the United States that could reshape not just OpenAI, the artificial intelligence (AI) firm behind ChatGPT they cofounded in 2015, but also the future of the technology more broadly.

Launched by Musk in 2024, the lawsuit is the culmination of a years-long feud that centres on the evolution of OpenAI from a non-profit to a for-profit enterprise.

The trial, which kicked off this week in California, is expected to last roughly three weeks. But its ripple effects could be felt for many years to come.

The case and the cast

The lawsuit pits Musk against Altman, OpenAI president Greg Brockman, OpenAI itself, and Microsoft, the AI firm’s largest backer.

Musk cofounded and helped fund OpenAI to the tune of about US$44 million. By his own account from the witness stand this week, he “came up with the idea, the name, recruited the key people, taught them everything I know, provided all of the initial funding”.

Brockman served as technical cofounder; Altman became chief executive in 2019. Their alliance with Musk fractured as the organisation grew. Musk departed the board in 2018. He says he was pushed out.

However, OpenAI says he walked when denied majority control. Musk subsequently launched his own rival AI venture, xAI, which is now part of SpaceX.

What Musk is alleging

As part of the lawsuit, Musk is alleging breach of contract, breach of fiduciary duty, false advertising and unfair business practices.

His core claim is that Altman and Brockman induced him to donate on the understanding that any artificial general intelligence – or AGI – built at OpenAI would stay “open” and shared with humanity.

Instead, Musk argues, the founders turned the charity into a “wealth machine”. They did this in two stages. First, via a 2019 capped-profit subsidiary. Here, OpenAI’s for-profit unit limited the returns, with the excess handed back to the nonprofit. Second, through a full restructure into a public benefit corporation, which is now valued at roughly US$852 billion.

Musk’s lawyers told jurors Altman and Brockman “stole a charity, full stop”. Outside court, Musk has been throwing insults at his opponents, prompting the judge to threaten a gag order.

OpenAI flatly rejects Musk’s narrative. As its lead counsel, William Savitt, told jurors:

We are here because Mr Musk didn’t get his way with OpenAI.

The company alleges, as described in two pre-trial blog posts, that Musk himself proposed merging OpenAI with Tesla in 2017 and walked away when denied majority control.

The lawsuit, OpenAI says, is “motivated by jealousy” and designed to damage a competitor.

A company under pressure

The trial arrives at a precarious moment for OpenAI.

The New Yorker magazine recently published an investigation describing Altman as a “pathological liar”. The investigation drew on an internal dossier compiled by OpenAI’s former chief scientist Ilya Sutskever which alleged a “consistent pattern of lying” to the company’s board.

Altman called the piece “incendiary” but acknowledged “a bunch of mistakes”. Musk has been amplifying the article to his X followers throughout the trial.

Financially, OpenAI is bleeding.

Internal projections point to roughly US$14 billion in losses for 2026 alone, with cumulative losses expected to top US$44 billion before any profit materialises.

Shortly before the trial began, OpenAI quietly shut down Sora, its flagship video-generation model.

Before closing, it burned around US$1 million a day in computing costs. The closure took down a US$1 billion Disney partnership with it.

Even a fresh US$122 billion fundraise from Amazon, Nvidia and SoftBank has not eased the pressure.

What Musk wants

Musk wants the jury to unwind OpenAI’s for-profit conversion, remove Altman from the nonprofit board, and strip both Altman and Brockman of their roles in the for-profit entity.

He is also demanding US$130 billion in damages from OpenAI – for what his team calls “ill-gotten gains”.

He has accused Microsoft of “aiding and abetting” and argues it is liable for a share.

His legal team argues OpenAI’s existing models already constitute AGI, because they have surpassed human intelligence in many tasks. Under the founding agreement, AGI could not be commercially licensed. This would include the licence currently used by Microsoft for CoPilot.

What’s at stake

If Musk wins, the consequences would be significant.

OpenAI’s planned initial public offering would almost certainly be derailed. This is expected in late 2026 at a US$1 trillion valuation. Investors in the recent funding round could face clawbacks.

Altman, the public face of the AI boom, could be removed from the company he has led since 2019. The broader question of whether AI labs founded as charities can lawfully pivot into commercial enterprises would be settled, at least in California. This has potential implications for Anthropic and other mission-driven peers.

Even a defeat for Musk would not end the controversy.

The trial has already pried open Silicon Valley’s normally sealed boardrooms, surfacing diaries, Slack threads and HR memos that paint an unflattering portrait of OpenAI’s governance.

The case crystallises a wider public anxiety: an incredibly powerful technology is being built and controlled by a tiny number of feuding tech bros. And it’s the rest of us who have to live with the consequences.

Rob Nicholls is a part of the University of Sydney Centre for AI, Trust, and Governance and receives funding from the Australian Research Council.