Director Aaron Sorkin, from left, Mikey Madison, and Jeremy Allen White pose for photographers upon arrival at the gala screening for the film "The Social Reckoning" in London, Tuesday, Sept. 29, 2026. (Photo by Scott A Garfitt/Invision/AP)

Is the Kids Online Safety Act the ‘Social Reckoning’ We Are Looking For?

A supporter of opposition presidential candidate Flávio Bolsonaro attends an Independence Day event the month before the general election in São Paulo, Monday, Sept. 7, 2026. (AP Photo/Andre Penner)

A supporter of opposition presidential candidate Flavio Bolsonaro attends an Independence Day event the month before the general election in São Paulo, Monday, Sept. 7, 2026. (AP Photo/Andre Penner)

On Eve of Brazil Elections, Meta Content Moderation Decisions Raise Questions

European Commission President Ursula von der Leyen delivers a speech at the annual State of the European Union at the European Parliament in Strasbourg, eastern France, Wednesday, Sept 16, 2026. (AP Photo/Pascal Bastien)

A Safer Internet Should Not Come with Democratic Shortcuts

Closed black iron gates adorned with golden wings and lanterns on each side in front of a grand university building surrounded by green grass.

Gates of the Royal Airforce College in Cranwell, UK in 2011. Adapted from 

Universities Need AI Infrastructure. They Don’t Need AI Lock-In

A hand is laid flat on the surface of a wooden classroom desk covered in children's scribbles and etchings, as a hand belonging to another person paints a blue ink mark on the index finger of the other hand. 

A voter puts his hand on a classroom desk as a polling officer applies indelible ink mark on his index finger at a polling station in a school in Kerala, India on April 26, 2024. (AP Photo/Manish Swarup)

A Misguided Tech Solution is Putting India's Electoral Integrity at Risk

What Counts as Scientific Research? The EU’s Digital Omnibus Could Set the Definition

A collage of images of hands with index fingers in pointing position and squiggly lines.

AI Transparency Starts With the Audience

Illustration from a screenshot of Meta founder and CEO Mark Zuckerberg’s July 14, 2026, deposition.

Illustration from a screenshot of Meta founder and CEO Mark Zuckerberg’s July 14, 2026, 

Eight Years On, Cambridge Analytica Scandal Catches Up With Meta in Santa Fe

Shattered black glass with some pixelated white and colored lights shining through.

Photo of a broken laptop screen, by Leon Brocard on 

Artificial or ‘Super Intelligent’? The Broken Metaphors of AI

Sam Altman (seated on top right end), CEO of OpenAI, arrives to speak at a UN Security Council meeting on Artificial Intelligence during the 81st United Nations General Assembly, United Nations Headquarters, NY, September 23, 2026. (Photo by Anthony Behar/Sipa USA)(Sipa via AP Images)

What Non-Western Governments Said About AI at the UN

A massive ship loaded with several dozen yellow trucks is docked by a port. A cement truck is lifted above the ground by a loading crane. The name of the ship is Chinafrie Happiness. This is written with white letters on the side of the ship. 

Chinese engineering vehicles are loaded onto a China-Africa liner at Yantai Port in Shandong Province for export to Africa on April 27, 2026. (Photo by Costfoto/NurPhoto via AP)

How China-Africa Relations Are Shaping the Geopolitics of AI

The US Capitol is seen Friday, Sept. 25, 2026, in Washington. (AP Photo/Mariam Zuhaib)

Google Ad Tech Remedies Show Need for Legislation

Prime Minister Anthony Albanese speaks to the media at Parliament House on September 8, 2026 in Canberra, Australia. The Albanese government is preparing new digital duty of care laws that would require digital platforms to take greater responsibility for preventing and addressing online harms. (Photo by Hilary Wardhaugh/Getty Images)

Australia Wants an Off Switch for 'The Algorithm.' That's the Wrong Fix

CEO of OpenAI Sam Altman speaks during a United Nations Security Council meeting on artificial intelligence and international security under “Maintenance of International Peace and Security” on September 23, 2026 in New York City. Tech industry leaders including the CEOs of OpenAI, Anthropic and Hugging Face briefed ambassadors on the risks of artificial intelligence. (Photo by Alexi J. Rosenfeld/Getty Images)

At the UN and in Washington, Leaders Clash on Approach to AI

People walk past posters encouraging participation in the 2020 Census, Wednesday, April 1, 2020, in Seattle's Capitol Hill neighborhood. (AP Photo/Ted S. Warren)

The 'Jenga Politics' Threatening the US Census

Activists dressed as queer versions of Gandalf, the character from Lord of the Rings, protest with a robotic pet Peter Thiel outside the headquarters of German publishing giant Axel Springer on September 24, 2026 in Berlin, Germany. Thiel, a Lord of the Rings enthusiast as well as an entrepreneur and investor who co-founded PayPal and Palantir, received an aware from Axel Springer. (Photo by Sean Gallup/Getty Images)

Gil Durán Maps the Rise of the 'Nerd Reich'

Maddie Freeman stands with her back turned to the side, in front of a billboard advertising her social media detox movement, with a deep blue sky in the background. The text on the billboard says "Get off your phone. Join our Free Tech Detox Movement. NoSoNovember.org". 

Maddie Freeman stands in front of a billboard advertising her social media detox movement, “No Social Media November”, in Denver, Colorado in 2005 (Photo credit: Maddie Freeman)

Meta’s Settlement Deal Won’t Give Me Back the Years I Lost to Social Media

Lord Tom Watson, former Labour deputy leader, speaks to the media on Wednesday January 22, 2025. (Press Association via AP Images)

Palantir's New Senior VP Swaps Lords Pass for Labour Leverage

Illustrative photo captured during the plenary session of October 2025 in Strasbourg, France.

“Behind the scenes - European Parliament in Strasbourg” by 

Europe’s Digital Sovereignty Agenda Must Include Digital Public Infrastructure

Thousands of people dressed in green and yellow rally on a street holding Brazilian flags and a giant balloon shaped like Donald Trump holding a smaller balloon shaped like President Lula wearing a prison uniform. 

Supporters of presidential candidate Flavio Bolsonaro with a President Trump inflatable, holding a doll of President Lula in a prison unifom, during an Independence Day rally in Sao Paulo, Brazil, Monday, Sept. 7, 2026. (AP Photo/Andre Penner)

Brazil’s Presidential Candidates Give Digital Sovereignty Many Meanings

A supporter of opposition presidential candidate Flávio Bolsonaro attends an Independence Day event the month before the general election in Sao Paulo, Monday, Sept. 7, 2026. (AP Photo/Andre Penner)

 Brazil President Luiz Inácio Lula da Silva’s campaign page from running political ads. The company 

 the restriction occurred because “some of its activities don’t follow our rules”, but did not specify which activities, or which rules. Lula’s Workers’ Party 

 with the company, and the restriction has since been lifted.

Lula is running for reelection against Flávio Bolsonaro, a senator and the eldest son of former President Jair Bolsonaro, who is now serving a prison sentence for an attempted coup d’état. “In practice, this is not only about the incalculable damage to the candidate and the elections, but the favoritism of his political opponents in the electoral process,” wrote the Workers’ Party campaign in its appeal to Meta. Tech Policy Press reached out to Meta for comment on the incident, but received no response by the time of publication.

The restriction on Lula’s campaign page followed another concerning content moderation decision by Meta on the same day. Last Wednesday morning, Meta took down an Instagram profile that had just the day before published a 

 about alleged connections between Flávio Bolsonaro and political scandals, including ties to a banker being 

 for the largest financial fraud in Brazilian history.

, which operates the Instagram profile that was taken down, is a project led by 

, a weekly podcast hosted by journalists Alessandra Orofino, Bruno Torturra and Gregorio Duvivier, together with two other independent news organizations.

Meta gave no explanation for the takedown, stating simply that the video violated its community guidelines, according to 

. After hours of public outcry and backlash against the company, the profile came back online in the middle of the afternoon. In response to an earlier set of questions sent by Tech Policy Press about this incident, Meta said only that “after internal review, the account mentioned in the report was restored. We apologize for the inconvenience.”

These two moderation decisions involving high profile candidates come amid a flurry of other concerning news about potential foreign interference in the elections by outside actors. Last Tuesday, the newspaper Folha de S. Paulo 

 that a group of right-wing influencers connected to United States President Donald Trump, including former Republican congressman Matt Gaetz, have been paying to promote posts showing Flávio Bolsonaro taking a lead in Polymarket bets. Although Polymarket is prohibited in Brazil, Bolsonaro and right-wing allies 

Two weeks ago, the Workers’ Party also requested an investigation from the Electoral Court on 

, a group of wealthy conservative donors co-founded by Vice President J.D. Vance, in Flávio Bolsonaro’s campaign. On the sidelines of the United Nations General Assembly last week, Trump 

 to designate 24 criminal groups, including two Brazilian factions, as foreign terrorist groups, as Flávio Bolsonaro had petitioned Trump to do. 

The action against Lula’s campaign page prompted many of his supporters to raise concerns around censorship. And one of the removed Instagram video’s authors called Meta’s decision—just 12 days before Brazilians head to polls to choose their next president—a 

 running for Congress, called it an act of foreign interference in Brazil’s electoral process.

The video posted on Instagram was a short teaser, linking to the 

 hosted on YouTube, which had reached the top of the ‘Hype’ ranking in less than 24 hours and was not removed. As of Thursday morning, the two teaser videos on Instagram had nearly 25 million views.

“The video script was the result of rigorous work from journalists, researchers, fact-checkers and a legal department,” Bruno Torturra, the editor at 

told Tech Policy Press. He added that the takedown was not preceded by any warnings, with a standard message sent after the profile had already been excluded.

“The video was not a campaign, it was not libel, not guesswork, it was a list of facts, all amply reported by the press and explained in an accessible format to everyone,” said Fábio Porchat, a comedian from the group 

“We showed the narrow relationships that exist between politicians like Flávio Bolsonaro, Nikolas Ferreira, the 

Baptist Church, Daniel Vorcaro and the Supreme Court of Justice,” he said. “After lots of pressure, the page came back online, but we don’t know for how long.”

In May, The Intercept Brasil published a bombshell investigation about 

, a banker currently under investigation for what has been dubbed as the biggest financial fraud of the country’s history. Since then, other media outlets have published 

 connecting other politicians, including federal deputy 

 to the banker. Many of these connections can be traced back to 

, a baptist church in the state of Minas Gerais.

 profile was taken down by Meta, there was outcry from Brazilian users, including artists, 

, with many calling Meta’s decision censorship. Many also raised questions about whether this episode can be considered foreign interference. The episode has sparked questions about sovereignty, freedom of expression, and the limits of big tech’s decisions amid electoral processes.

“Beyond the story told in the video, this case reveals another scandal: Meta’s arbitrary, opaque, and disproportionate power over the national media landscape”, said Torturra. “It also highlights a central issue for the State and society to consider: can we speak of freedom of expression on social media without algorithmic transparency? Is the press truly free when subject to unwritten rules? To codes over which we have absolutely no editorial control or influence? To the whims of nameless executives?” he added.

Torturra said the group is now working on a script for an upcoming video about platforms, digital sovereignty and the privatization of freedom of expression.

Last Wednesday afternoon, Ctrl+Z, a Brazilian nonprofit organization focused on big tech accountability, 

 the Federal Prosecution Service to investigate Meta and demand the company detail the reasons behind the removal of the Instagram account. (Tatiana Dias, a program director at Ctrl+Z, is a fellow at Tech Policy Press.) The case was distributed to the Regional Prosecutor's Office for Citizens' Rights in São Paulo and a prosecutor has been assigned to it, according to confirmation from the press office.

“The removal of an established journalistic profile—with a track record, an audience, and editorial partnerships—without specifying the particular rule allegedly violated runs counter to the duty of transparency and due process that the Supreme Federal Court’s own jurisprudence has come to require of platforms, and it frustrates the right of the user and the public to know and contest the grounds for such a severe measure,” Ctrl+Z wrote in its request.

In addition to Ctrl+Z, federal deputy Orlando Silva, the former rapporteur of Brazil’s 

 the Federal Prosecution Service investigate Meta over its decision to shut down the profile.

 profile is not an isolated event, explained Luã Cruz, director of strategic litigation at Ctrl+Z. Over the past few months, the organization has been tracking cases of censorship to newsrooms and individual journalists.

“We understand this is another action by Meta like others that have been happening, and more strongly after Trump took power, after Zuckerberg changed Meta’s content moderation policies, with the use of AI, and the lack of explanation to users who have their accounts taken down or banned,” said Cruz.

If this weren’t a case involving famous actors, with wide media coverage, it would likely fall “into the limbo,” like many other businesses, he pointed out.

“We can't have this ‘black box’ situation where content of such great importance goes offline—without us knowing why—and then comes back online—again, without us knowing why, or how it returned, what the decision or reason was, or what the process was that caused the account to be taken down and subsequently restored,” he added.

(2010), comes out in theaters this October. The film dramatizes the 2021 release of the 

, which exposed the company’s research and neglect of teen mental health and the spread of misinformation on its platforms. The sequel arrives in a world almost unrecognizable from the one Fincher depicted in 2010. With scientific research still uncovering 

 between compulsive social media use and declining mental health, there’s a growing consensus that social media poses a serious danger to minors. 

 reported this past July that nearly 6 out of 10 adults in the US support a social media ban for those under 16, and 78% support requiring people to verify their age before using these platforms. And on Aug. 26, Meta settled a 

 over claims that the company intentionally designed features on Instagram and Facebook to foster addictive behavior in teens and misled the public about the safety of its platforms. With this momentum and public support, a national law regulating social media platforms has never looked more likely.

Two bipartisan bills are under consideration. The House of Representatives 

 the Kids Internet and Digital Safety Act (

) in June. The Senate Commerce, Science, and Transportation Committee advanced the Kids Online Safety Act (

) in August. The full Senate has not voted on KOSA, and so the two chambers have yet to reconcile their approaches. Senator Marsha Blackburn (R-Tenn.), a co-sponsor of the Senate bill, has been clear that she finds the House version 

: “It is toothless, and it is a very pale imitation of the Senate version, and therefore it is a pale imitation of big tech accountability.”

The Senate KOSA supporters are shrewdly capitalizing on the public backlash against companies like Meta for “hacking our brains” to sell our information to advertisers, and marketing the regulation as a means to mandate safety by design measures across the industry. Some nuance missing from the coverage of this recent surge in social media regulation is that the bill does propose some real changes. But new research suggests it overlooks the most critical levers available to create safer online spaces, like the structure of the platforms themselves, while unleashing a whole new set of issues that arise with age verification and exposing platforms to legal liability.

KOSA’s aim is to ensure that online platforms prevent and mitigate specific dangers to minors, including eating disorders, depression, anxiety, online harassment, sexual exploitation, and compulsive use (Sec. 102(a)). For users the platform knows are minors, it would require the most protective privacy and safety settings by default, limiting design features such as infinite scroll and autoplay, and giving minors an option to opt out of personalized recommendation systems (Sec. 103(a)(1), (3)).

Social media regulations have up to this point attempted to target different aspects of digital platforms like the content published by users, the algorithm, and now these shiny trinkets that keep users entranced, like rewards for Snapchat streaks and autoplay. The platform’s architecture, on the other hand, is not any single feature but the structure which governs the extent to which users can connect and what actions are permitted. This connective tissue differs drastically between different online platforms and plays a large role in shaping what content can go viral and reach minors.

To help clarify that distinction, consider research from George Washington University Professors David Broniatowski and Lorien Abroms, Principal Researchers at the GW Institute for Data, Democracy, and Politics (IDDP). Their 

 demonstrates how remedies like content moderation and targeting specific features can produce mixed results depending on a platform’s structure.

Studying the spread of misinformation during the COVID-19 pandemic, Broniatowski and his colleagues at the University of Maryland found through a 

, still under peer review, that the effects of shifting from a popularity-based algorithm to a chronological one can vary substantially by platform. On Facebook and Twitter, changing the algorithm produced a modest shift in content quality. In contrast, on a platform like TikTok, where any content can reach any user, the model estimated that the popularity-based algorithm reduced the average informativeness of the content users saw by 92%, because it promoted whatever happened to gain early traction. Reddit, organized by walled subreddit pages and moderators, had virtually no effect from changing the algorithm type.

Separate studies by Broniatowski, Abroms, and their colleagues observed that 

 attempts to remove anti-vaccine posts and profiles increased the virality and political polarization of the content and accounts that remained. The architecture of these platforms (Facebook’s linked pages and groups; Twitter’s unrestrained retweet network) permitted motivated users to construct new routes to anti-vaccine content once old ones were removed. Groups and retweeting networks could simply reconnect severed clusters. In other words, the types of actors on a platform (moderators, groups, followers) and how they can interact determined whether the interventions reduced negative exposure.

Advocacy groups have raised other important shortcomings including KOSA’s “teeth,” its duty of care provision, which holds platforms liable for their design features contributing to clinically diagnosable symptoms (Sec. 102, 109). 

 notably excludes the duty of care standard, citing 

 over the impact on users’ freedom of speech. In response to these critiques, KOSA’s Senate sponsors have been 

 that the bill “does not give the FTC or state Attorneys General the power to bring lawsuits over content or speech.” They argue they are targeting the product features that are designed to keep minors hooked, not regulating the content that is posted on the platforms.

How companies may comply tells a different story. Although KOSA doesn’t directly regulate speech, design obligations can affect how platforms moderate, organize, and rank content. 

 Kate Ruane, director of the Free Expression Project at the Center for Democracy & Technology, KOSA 

 would require the companies to proactively reduce children’s exposure to content that can be deemed harmful. The legal exposure could incentivize companies to overcorrect, 

. There’s precedent for the concern. Social media platforms reacting to the Fight Online Sex Trafficking Act and the Stop Enabling Sex Traffickers Act (FOSTA/SESTA) 

 non-explicit content sometimes based on how much revenue the accounts generated and the user’s body type, which especially policed Black women’s bodies. A report from 

 also found that 92% of the leading parental control apps blocked sites led by LGBTQ+ organizations or that provide sexual education. And as recently as 

, Meta’s teen accounts restricted posts with hashtags like #queer, #transgender, or #lesbian, flagging them as unsafe for months.

One consequence, or arguably a necessity, of creating a safe space for minors online is that it requires correctly identifying who is an adult or the parent of a minor. KOSA does not 

companies should identify minors on the platform, but the legal exposure pressures companies to instill some method that is not easily circumventable. This can include requiring all users to provide either biometric data like a face scan or upload proof of age documents, threatening the anonymity users have enjoyed on the internet for decades.

As Santa Clara University law professor Eric Goldman shares in 

The “Segregate-and-Suppress” Approach to Regulating Child Safety Online

, age authentication increases the vulnerability of sensitive data substantially. A new norm with age verification also puts people, and minors especially, in a position where they constantly have to decide whether they should share sensitive information online, and poses a great opportunity for bogus sites designed to extract authentication data. By not formally imposing an age verification regime, KOSA doesn’t have to address these 

This is not to say that reducing children’s exposure to features like infinite scroll is senseless. Rather, all users, whether they are 14, 22, or 45, should be empowered to opt out of a more engagement driven and addictive experience. KOSA takes steps to expand that agency but still operates a form of governance where safety (and plausible deniability) is dependent on knowing who is a child, pressuring platforms to make broad editorial decisions over algorithms. Researchers like Dr. Broniatowski suggest one 

 is to establish social media building codes that truly embody a “safety by design” approach. Such codes could discourage certain features, including late night notifications and infinite scroll, while requiring platforms to disclose how their structure and algorithms can cascade problematic content and how they seek to mitigate those effects.

KOSA has undergone several iterations to appease its critics, and this might be the year its opposition finally relents. But the vigorous efforts to protect minors can just as well be redirected toward bringing about a tangible change to the status quo for all users by adopting a structural approach that treats social media as the public forums they are. A true social reckoning calls for standards that prioritize safety over engagement, without sacrificing anyone’s privacy, agency, or free expression.

 address by European Commission President Ursula von der Leyen drew a bated breath for many in the digital rights community, as we awaited the announcement of the 

, new legislation aimed at creating a safer online environment for minors.

The online space offers young people the chance for connection and exploration, but it also exposes them to significant risks of 

, among other recognized harms. The proposal is a response to wide-reaching calls from Member States and the general public for 

 to address these well-documented online harms, but the devil will be in the details.

As a new addition to the EU's digital rulebook, the proposal also arrives amidst growing institutional pushback against, and in some cases dismantling of, fundamental rights protections in some of the EU’s defining digital rules.

A worrying democratic backsliding on key digital files

The ink was barely dry on the EU AI Act before the 

 was introduced and pushed through the legislative process under an expedited procedure. The projected ‘tension’ between enforcing risk-based rules and championing innovation placed key fundamental rights protections first on the chopping block.

Civil society organizations successfully pushed back on the worst proposed changes, but many aspects were still 

. The central pillar of data protection rules — the General Data Protection Regulation (GDPR) — risks having key provisions on personal data be undermined through 

changes on how highly technical concepts such as pseudonymization should be interpreted

 and a wider mandate for AI-related data processing.

Most publicly, the European Parliament President enacted an 

urgent procedure to pass a vote on the lapsed e-privacy derogation

, commonly known as Chat Control. The Parliament had previously voted down the framework twice because of the prevailing rights concerns many civil society groups 

. This wave of procedural backsliding has led many to hold a pessimistic view of when and how best public interest advocates can continue to push to ensure that the EU’s evolving digital rules uphold our fundamental rights.

The concern is that, despite this welcome step forward in addressing online safety for minors, this new proposal falls into the same pattern of EU tech policies being shaped outside democratic due process. Amid the European Commission’s push for simplification of digital rules, procedural mechanisms to ensure transparent and multi-stakeholder engagement in policymaking are being undermined, and the adoption of the Act is no different.

Although debate on child safety online has been ongoing for months, and several member states have taken matters into their own hands by introducing a 

, it can be argued that the Commission had little choice but to propose the new KIDS Act.

The Commission President took the diligent step of establishing a special panel to develop a report with recommendations for EU action, informed by notable experts in the field. Despite many of the 

 of the report being embedded throughout the proposal, this does not negate the lack of impact assessment and stakeholder consultation, which are essential requirements of the 

 of a similar law in France should have reinforced the need for diligent assessment and stakeholder engagement, not justified sidestepping these procedures.

This is all the more necessary given that the proposal explicitly aims to turn the 

Digital Services Act’s Article 28 Guidelines

 into ‘hard law’ through this new law, only strengthening the argument that the extensive consultation process that led to the development of the guidelines should have been mirrored in the knowledge that these non-binding measures are to become binding legal obligations. This embrace of procedural sidestepping is all the more concerning given the wide-ranging fundamental rights implications of the proposal.

Good intentions, but fundamental rights must be balanced

For many civil society groups, the proposal has caused mixed reactions. On the surface, the fact that the European Commission is not proposing a blanket social media ban is appreciated. As conveyed by the 

, bans and age restrictions are blunt legislative tools that fail to address the core issue of platform safety. 

The proposal, which is highly complex, incorporates safety-by-design measures that many advocates have 

 to addressing the online harms that not only young people, but all users and democracy at large, continue to face. Still, the draft proposal includes far-reaching age verification and age assurance requirements which undermine this more measured approach.

Advocates have long stressed the concerns for many existing age assurance tools, outlining a more 

 for lawmakers to consider. At the heart of the European Commission’s proposal is a reliance on the EU’s age verification scheme, with zero-knowledge proof presented as the privacy-preserving gold standard. 

, however, that zero-knowledge proof is not a silver-bullet solution and has its 

. The proposal also envisages parental oversight via an app, which itself will require confirmation of guardianship, thereby adding another layer of ID verification for parents.

From a free expression perspective, there are questions as to whether the proposal respects the right to free expression and access to information for children under the age of 13, 

which are rights they maintain and should still safely enjoy

 about an age-gated internet, especially for users from marginalized communities. For teens aged 13-15, ‘mini’ accounts with limited features are available, but parents can monitor account activity. Whether teens can freely enjoy these rights under the premise of parental monitoring is also a concern. Whether parents generally support this approach, given 

 preference for more personal choice, remains to be seen.

In short, the solution on the table, despite having some welcome aspects, may unintentionally undermine the fundamental rights of not only children, but all online users. Given this, concerns about legislative shortcuts and expedited procedures, where opportunities for public debate and expert consultation are reduced, are simply unacceptable; the issues this law aims to address and its potential impact are simply far too important.

Getting the process right: democratic oversight must come first

As this law moves forward, public consultation and democratic scrutiny will need to be forefront. Children and young people, expert civil society and public interest technologists must be extensively consulted if this law is going to land where we hope it eventually will; as a complementary framework to laws like the Digital Services Act and the upcoming Digital Fairness Act, that truly protects all communities online.

Strengthening measures to ensure platforms meaningfully mitigate the harms they have thus far failed to address, while respecting the privacy and expression rights of children and all users, is critical. As the 

 “Europe has the power to act, and it is we who decide our rules.

” The EU also has an obligation to establish these rules through an open, democratic process, informed by relevant expertise.

Universities are moving rapidly from debating artificial intelligence to buying it. That shift is understandable. Students already use generative AI extensively, employers increasingly expect graduates to understand it, and leaving access to the best systems dependent on who can afford which commercial subscriptions creates obvious problems of imbalance.

In the United Kingdom, Universities UK has now 

 for AI tools to be accessible to every undergraduate student. In Australia, UNSW Sydney has 

 one of the largest higher education deployments in the Asia-Pacific region, providing ChatGPT Edu to more than 80,000 students and staff.

These initiatives point toward a future in which institutional access to AI becomes as normal as access to a learning management system, academic databases or university email.

Universities do need AI infrastructure. But they do not need AI lock-in.

That distinction is becoming urgent because universities risk treating AI procurement as though they were buying another piece of software. Generative AI is developing into something considerably more consequential: a computational layer through which students learn, academics research and professional staff increasingly perform institutional work.

A procurement decision can gradually become an institutional architecture.

UNESCO put this problem unusually clearly on the international policy agenda in September. At its 

 in Paris, education ministers called for AI systems used in education to be auditable, for data to remain portable and for education systems to consider the total costs of ownership before adopting AI. The resulting ministerial 

 explicitly favors interoperability, portability, and open-source systems so that education systems retain the ability to change course.

The important principle can be expressed more simply: institutions need to be able to leave.

This is not an argument against commercial AI providers. Universities need access to capable systems, and institutionally managed versions of commercial products can provide significant advantages over consumer accounts in security, privacy, administration and equitable access.

The problem comes when access becomes dependence. Consider what happens over several years after a university selects a preferred AI platform. Staff develop workflows around it. Students learn through it. Bespoke assistants are created inside it. Institutional knowledge accumulates within it. Other university systems are connected to it. Eventually, AI agents may begin initiating actions across those systems.

At that point, changing providers is no longer comparable to replacing a software license. The institution must unwind an ecosystem. This is a familiar feature of 

: the value of a platform increases as more activity takes place within it, while the cost of leaving rises at the same time.

 with learning management systems, cloud infrastructure, academic publishing and research analytics. AI potentially intensifies it because the technology is not simply storing information or providing access to content. Increasingly, it is mediating the work itself.

The relevant procurement question is therefore changing. It is no longer only, “which AI system offers the best capabilities today?” Universities also need to ask, “what architecture preserves our ability to choose differently tomorrow?”

The university should control the layer above the model

UNSW offers an interesting glimpse of one possible answer. Alongside its large OpenAI deployment, the university says it is developing a multi-model AI environment in which students and staff can experiment with different AI systems as the technology evolves. That may ultimately prove more strategically important than the ChatGPT agreement itself.

The objective for universities should be to separate their institutional AI environment from any individual model operating within it.

An institutionally governed layer could establish common rules for identity, data access, security, privacy, and appropriate use while allowing different AI models to operate underneath. A university might use one model for particular research tasks, another for coding, and another for routine administrative work. Those choices could change as technology, cost, and evidence change.

Today’s leading model may not be tomorrow’s. Universities should be positioned to benefit from competition between providers rather than finding themselves locked into the winner of a procurement exercise conducted several years earlier.

This principle matters beyond higher education. Public institutions are under increasing pressure to adopt AI while a small number of companies control many of the most capable models and the cloud infrastructure on which they depend. Will Stronge, chief executive at the Autonomy Institute, has recently pushed the argument further, describing the AI systems through which students increasingly think and research as a form of “

” and arguing that questions of ownership and control therefore belong at the heart of educational policy.

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe