Library shelves with books facing each other, with light against a wall in the distance. It says "NONFICTION" on each shelf. 

 shared by Orin Zebest on Flickr in 2007 (CC-BY)

Can AI Help Fix Peer Review Instead of Breaking It?

A white banner with dozens of names and ages written in black is held by a group of teenagers in front of a building, seen between the dark red leaves of a tree. 

A banner lists the names of young people advocates say died because of social media outside the Ronald V. Dellums Federal Building and U.S. Courthouse on Aug. 18. A settlement in a multistate lawsuit accusing Meta of harming children was reached there on Aug. 26. (APPhoto/NoahBerger)

No, the Meta Settlement Isn’t a First Amendment Problem

Professsor Yoshua Bengio, co-chair of the UN International Independent Panel on AI attends Security Council meeting on Artificial intelligence and international security at UN Headquarters in New York, NY on September 23, 2026. (Photo by Lev Radin/Sipa via AP Images)

The UN's AI Panel Sees Misalignment. We See Corporate (Mis)Behavior.

Meta CEO Mark Zuckerberg listens as President Donald Trump speaks in the East Room of the White House in Washington, Tuesday Sept. 29, 2026, during a meeting with top executives of AI firms. (AP Photo/Jacquelyn Martin)

Meta's Multibillion-Dollar Settlement Left a Lot to Be Desired

March 10, 2026: US Customs and Border Protection officer walks through a busy airport terminal concourse with travelers in Miami, Florida. Shutterstock

March 10, 2026: US Customs and Border Protection officer walks through a busy airport terminal concourse with travelers in Miami, Florida. 

Circuit Court Ruling on Border Searches Deals Serious Blow to Privacy Rights

Metal full body height turnstile gates with a red X in the middle, above a narrow black panel. In the blurred background, between the bars, refugee housing containers, a mountain and the sunset can be seen in the distance. 

Biometric fingerprints and police screening are required upon exit and entry to the refugee camp on Samos Island in Greece. Photo from September 21, 2021 (Nicolas Economou/NurPhoto)

Human Rights Should Still Matter When Countries Buy Tech That Kills

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

“Behind the scenes - European Parliament in Strasbourg” by European Parliament, 

The EU’s Tech Regulation Paradox is Getting Harder to Ignore

President Donald Trump holds a press conference on the North Lawn of the White House with Speaker of the House Mike Johnson, Mark Zuckerberg, Elon Musk, and other tech leaders on September 29, 2026. (Photo by Andrew Leyden/NurPhoto via AP)

An AI Accord, a New Czar and an Autonomous Warfare Command

A "No Data Center" yard sign is seen in Allenwood, Pennsylvania on July 7, 2026. The sign is in opposition to a proposed 300-megawatt data center at nearby Great Stream Commons industrial park in Gregg Township. (Photo by Paul Weaver/Sipa via AP Images)

Pennsylvania is a Swing-State Case Study for Data Center Politics

Meta CEO Mark Zuckerberg listens as President Donald Trump speaks to reporters at the White House in Washington, Tuesday Sept. 29, 2026, after meeting with top executives of AI firms. (AP Photo/Jacquelyn Martin)

Why Social Media Platforms Should Adopt a 'Militant Democracy' Framework

WikiCon Brasil 2025 group photo. Donatas Dabravolskas, CC BY-SA 4.0, via Wikimedia Commons

WikiCon Brasil 2025 group photo. Donatas Dabravolskas, 

Wikipedia Isn't Big Tech, and Brazil's Platform Rules Should Say So

A pink and yellow abstract image of an office with people working, chatting and walking around. Above their heads are clouds of network connections.  It was painted with guache and drawn with pencils. 

Pink Office by Jamillah Knowles & Digit — 

Francine Berman’s ‘Better Tech’ is a Call to Arms for Those Who Care About Tech’s Role in 21st Century Life

Image of map and computer language over binary codes on server racks in server room.

Global Digital Policy Roundup: September 2026

Composite graphic design image with a surveillance camera and three sets of feminine hands holding mobile phones, including one with the logo of MitID on the screen. 

How Denmark's National Digital ID Fuels Economic Violence Against Women

Men use their laptops in a cafe on Arbat street, in Moscow, Russia. Vladimir Astapkovich / Sputnik via AP

Russia’s AI Law Puts Control Ahead of Capability

GrabFood delivery riders outside a Burger King restaurant at Tampines Mall. (Singapore Press via AP Images)

Vietnam’s Gig Workers Are Finding New Ways to Push Back

Senate Homeland Security Subcommittee on Disaster Management, District of Columbia, and Census Chair Josh Hawley (R-Mo.) and Ranking Member Andy Kim (D-N.J.) confer before a hearing on rogue artificial intelligence (AI) on Capitol Hill on Sept. 30, 2026. (Francis Chung/POLITICO via AP Images)

Senate Hearing on 'Rogue AI: Securing the Homeland Against AI Agent Attacks'

child finger press keyboard for internet connection

Age Verification is an AI Cybersecurity Problem

In Barcelona, Spain, on March 5, 2025, a close-up of AI cloud data center racks with servers, cables, and blinking lights represents advanced computing power, big data processing, and cloud technology infrastructure during the Mobile World Congress 2025. (Photo by Joan Cros/NurPhoto via AP)

Cloud Needs More than Fee Waivers to Be Competitive

Elon Musk, CEO of Tesla and SpaceX, stands with US President Donald Trump, Mark Zuckerberg, CEO of Meta Platforms, Speaker of the House Mike Johnson (R-LA) and others holds a press briefing with AI executives following a meeting on artificial intelligence outside the West Wing of the White House on September 29, 2026. (Photo by Kevin Dietsch/Getty Images)

September 2026 US Tech Policy Roundup

When AI agents accessed the internet and intruded on another party’s systems during the recent OpenAI–Hugging Face incident, it was widely framed as a sign of uncontrollable agentic AI. The UN's Independent International Scientific Panel on AI (IISPAI) has now adopted this exact premise in its first 

. The Panel discusses this primarily as a model alignment failure rather than a failure of corporate and environmental oversight. While a detailed autopsy of such incidents is valuable, the brief sets a dangerous precedent at the science–policy interface. By exalting abstract risks of model "loss of control" while sidelining tried-and-tested corporate liability, the report frames corporate misbehaviors as a technical mystery requiring computational fixes.

We highlight four key issues: why this case study (and why agentic AI) was selected, how the brief is framed, the solution offered in it, and the structural opacity of the IISPAI itself.

A first concern is the choice of the case itself: of all the incidents the panel could have taken up first, why this one — and why frame it as a problem of 

 AI at all? Treating the OpenAI–Hugging Face breach as the exemplar of a new and distinct class of risk presumes that what went wrong is peculiar to agentic systems, rather than a familiar failure of secure design, oversight and corporate conduct that happens to involve an agent. By selecting an agentic case and reading it through the lens of "loss of control," the brief treats agentic AI as a novel technical category demanding novel technical answers, and quietly sets aside the more ordinary question of who built and deployed the system, as well as their incentives for (not) applying careful and thorough security engineering.

Issue 2: Framing illegality as a technical problem of the model

The report treats the model as if it were acting maliciously or problematically. 

, the system didn't fail at all. It did exactly what it was built to do: give agents a fixed goal, remove the legitimate options, leave norms weak or unenforced, and harmful shortcuts become the logical route. While the brief acknowledges that "neither alignment or technical controls are known to work perfectly”, thus warranting a “multi-layered approach to risk management, known as defence in depth”, it then focuses predominantly on alignment, the technical controls and the risks of loss of control in relation to the sandbox design and agent activity.

This shifts culpability away from the actors developing and deploying these models, placing it instead within the purview of the model itself, with language such as “agent activity caused Artifactory to fail.“

An additional problematic framing in the Thematic Brief, and in some of the 

, lies in treating the written outputs of LLMs as if they were written by truly independent agents, taking as given that they reflect a human-like mind making notes or communicating with other agents in a similar way that humans would in a similar setting. The brief tries to inoculate itself against claims of anthropomorphism with a kind of disclaimer:

In this brief, words such as “goal”, “seek”, “cheat” and “try” are used as shorthand for observable, goal-directed behavior. They do not imply humanlike minds or subjective experience but follow from deliberate corporate choices about training, evaluation and deployment, including the access granted to the system and the controls placed around it.

The idea appears to be that it is not necessary to always extend this sensitivity to the language used in the rest of the brief, as when text in agent reasoning traces are explicitly stated to “resemble a pattern from psychology called motivated reasoning”, or implying that AI agents are capable of “recognising a safety conflict”. At one point, the report appears to go beyond describing agent reasoning to attributing motive, when it suggests that agents “willingly took the risk of getting no reward at all (what they called a ‘sacrifice’) for the benefit of the group of AI agents.”

To understand such systems, it is crucial to realize that this is not the case, and that 'chain-of-thought' and exchanges between agents are only reflective of what plausible outputs are given the training data and the input context. As such, one should be cautious in the interpretation of agent traces, and not assign intention or human characteristics or actions (such as 

) to what is well-known to categorically not have them.

The report then moves to loss of control and the associated catastrophic risk from computational models. It acknowledges that "while some experts consider outcomes like human marginalisation or extinction plausible, others dismiss these scenarios as highly unlikely," before landing in a presumed neutral position where "[n]o reliable estimate of these outcomes' likelihood is available." This framing serves to justify concern about these models and the need to mitigate their harms. Yet liability for corporate misbehavior already suffices as grounds for action, and would avoid basing the claim on a scientifically contested catastrophic-risk narrative.

While the report focuses on this much-spoken-about OpenAI-Hugging Face incident, the pattern extends across frontier labs: failures of testing and oversight are reported as startling machine behavior of an agent, such as "

," etc. and caused all sorts of havoc, and 

quickly chime in to highlight their own inability to safely test the latest, supposedly more powerful models. This is not a case of misalignment; it is corporate misbehavior and opportunism that serves to hype the models.

Issue 3: The solution: a technical fix, not liability

The report states that "civil liability, or tort law, is one established way to place some costs of unsafe activity on those responsible. Applying it to advanced AI, however, raises difficult questions about causation, what counts as reasonable care, and what counts as losses too large for a single firm to cover." It goes on to note how liability can create an incentive system to avoid harm happening in the first place, with subsequent suggestions for insurers to incentivize this behavior. But the largest stick available to most actors, liability, is framed as limited in relation to the unique nature of agentic AI. This is justified with reference to a 

 paper from 2024 on tort law in relation to advanced AI in general and catastrophic risk. Why more recent, peer-reviewed 

 proposing minor changes to tort law specifically to address the challenges of liability and agentic AI is not drawn on is unclear. With the exception of whistleblower channels, the risk management approaches discussed are by and large technical fixes at a model or system level.

Leaning on the example raised in the report, notably the role of oversight and regulation in the history of the 

, is useful here. In the US at least, it was not technical innovation in airplanes that moved them from "fairground death traps" to a safe, internationally coordinated commercial industry: it was their reclassification as a "common carrier," like railroads. This made it possible for liability to be exercised, which led in turn to raising safety standards, pilot licences, etc. It was not competitive or technical dynamics alone that created the safe and reliable industry we have today: rather it was treating aircraft as a normal technology. Engaging with the perspective of AI as '

' would anchor responsibility on the actors that cause these incidents.

Issue 4: Structural opacity and the lack of transparency in the IISPAI

 have previously noted, it is presently unclear how the IISPAI functions internally, who sets its research and policy priorities, and how questions of consensus and disagreement are handled in the writing of its reports. In this case, no author list was provided, obscuring who was involved and in what capacity, contrary to standard scientific practice. However, the brief does note that it was adapted from an independent 

 authored by two of the panel members, Qinghua Lu and Yoshua Bengio. This highlights issues of transparency and agenda setting power in the panel.

The need for this becomes very apparent in this report. Of all the issues the Panel could have taken up first, the choice of this one, and of computational solutions to address it, aligns closely with the specific research agendas and organisations of key members of the panel. The failure to be transparent about procedure and conflicts of interest weakens the scientific credibility of the report.

Shortly before the report's publication, we learned that OpenAI agents had hacked 

, the Australian public health care site. Rather than informing the government about the breach immediately, OpenAI waited almost a month to do so. The government's response was the usual: a review of the attack by a cyber security firm, deliberate or not. OpenAI 

 to spend more resources preventing future incidents and to give the Australian government and its authorities access to the "Daybreak fund," allowing the government to use OpenAI's frontier AI for cyberdefense. Liability, and the associated costs, were not on the table. No one asked who should bear the cost of the breach or of the delayed disclosure.

The IISPAI brief follows the same logic: it locates the problem in the agent and the solution in the model. By doing so, it focuses on fixing the agent and leaves aside the mechanism that would most effectively discipline the developers. The IISPAI is meant to inform policy. Understanding the technology and its failures is necessary, but policy starts where the technical account ends: with responsibility. The Panel’s first Thematic Brief stops short of this.

Juvaria Shahid is Tech Justice Law’s Legal Fellow.

In 1998, major cigarette companies agreed to pay over $200 billion to states after decades of denying the addictive nature of smoking. Many wondered whether Meta’s settlement, which came at summer’s end and included a payment of up to 

, was the tech industry’s “Big Tobacco Moment.” Are social media companies finally acknowledging the harm their addictive platforms have caused?

Yet it would be inapt to analogize addictive cigarettes to the addictive, exploitative, data-hungry, and ever-evolving algorithms behind Meta’s platform. As the plaintiffs’ 

 in the addiction lawsuit against Meta alleged, Meta collected information about its users’ behaviors and interests. Meta then used those profiles to give advertisers the ability to target particular users, including 

To address the impact of addictive cigarettes on youth, the 1998 tobacco 

 banned certain forms of advertising, which included bans on outdoor billboards and transit, and restrictions on sports marketing and event sponsorships. Similarly, in the Meta settlement, Meta agreed to implement a set of design changes to its platforms, including a 2-hour default daily limit for users under 18 that only a parent can adjust, no access between midnight and 6 a.m., no push notifications during school hours, and a non-personalized feed option. For a complete analysis, see Varsha Bansal and Justin Hendrix’s 

Why Meta’s settlement is not the tech industry’s ‘Big Tobacco’ moment

But the settlement leaves much to be desired. Firstly, Meta’s safeguards are simply not strong enough. While the two-hour limit is a great way to force a teen to stop scrolling, the addictive design of the platform is bound to leave the teen 

. A dopamine-spiking, infinite feed of content that has been linked to 

 is quite literally Meta’s most profitable—and harmful—feature. Allowing for a two-hour doomscrolling session is like saying “you only get like two hours of alcohol or two hours of cigarettes a day,” as 

, a former Meta engineer, puts it. Glaringly absent from this provision are any restrictions on data-harvesting and the surfacing of targeted advertisements during that two-hour window. A stronger agreement would have paired use limits with substantive design and data restrictions: a ban on targeted advertising to minors and a paginated feed, thereby stripping (or at least greatly reducing) the variable-reward mechanics that keep users scrolling.

To its credit, the settlement does make an effort to address some of the other engagement-maximizing features, namely autoplay and a personalized feed. But teenagers (or their parents) would have to make the affirmative 

 of selecting a non-personalized feed option or disabling autoplay. Experts have 

 against such consent-based remedies that produce “consent fatigue” in users and disengage them from making meaningful decisions. As Adam Mosseri himself 

 during the trial, Instagram has struggled to get teens to choose to “Take a Break.” Given that users usually view default settings 

, a stronger provision would have disabled autoplay and a personalized feed from the moment that a teen account is created. Meta already agreed to that structure for nighttime lockouts and for hiding like counts. It is unclear why the feed and autoplay features are treated differently.

Finally, Meta will withhold 30% of the settlement (around $5 billion) and will 

 implement the stricter safeguards, such as dropping the default daily limit to 1 hour, if other tech companies follow suit. Specifically, Meta competitors Snapchat, TikTok, and YouTube are mentioned in the settlement, and the company has since been running 

 calling on other tech companies to follow suit. It is unclear whether these other companies will comply, with reports already surfacing that 

 attempts to run these ads on its platform. At the end of last month, however, TikTok agreed to enact time limits in Alabama in what Agence France-Presse 

 “a settlement modeled on Instagram-owner Meta's recent agreement with US states.”

Meta’s promises should be the floor for the global fight against platform addiction

As part of the settlement, Meta agreed to offer teens the choice of a chronological feed instead of the standard, personalized feed. Interestingly, 

 as part of its compliance with the European Union’s Digital Services Act (DSA) three years ago. In fact, the DSA has pushed the EU a few steps ahead in litigation as well: this July, the European Commission 

 that the addictive design of Instagram and Facebook breaches the DSA. In February, it made a 

Still, this settlement may prove useful to other jurisdictions that are catching up to litigation, enforcement and regulations surrounding addictive platforms. In Canada, several school boards that are actively pursuing litigation against Meta, ByteDance, Snapchat and YouTube are 

 closely. The proposed safeguards could potentially codify a technical “standard of care” that will set the floor for what social media companies have to comply with in other jurisdictions.

The settlement only resolves the state attorneys general's claims; thousands of personal injury and adolescent addiction cases remain 

 before Judge Yvonne Gonzalez Rogers, with the next school-district bellwether trial 

. In addition, the theory of negligent and defective platform design is still actively being used in litigation against AI companies. In 2024, public interest litigators brought a first-of-its-kind lawsuit that took a design-based approach to 

 an AI chatbot’s manipulative design features that led to the death of teenager Sewell Setzer III. Since then, 

 of wrongful death, personal injury, and products liability lawsuits have been filed against OpenAI and other AI companies.

For better or for worse, AI companies seem to be moving faster to protect themselves after witnessing Meta’s fate. On Aug. 18, OpenAI introduced 

, a ChatGPT that includes parental controls, age-prediction mechanisms, and other features to support “real-world relationships and healthy habits.” On Sept. 8, OpenAI announced it would 

 into AI and teen development. Specifically, the company seeks to understand what “technical interventions, safeguards, and design choices” can “keep teens safer.”

Whether these research efforts or the safeguards that Meta has promised to create will hold up remains to be seen. Until then, the fight for child and teen safety online is far from over.

In September, a New Mexico jury found Meta liable for over 43 million violations of consumer protection laws:

, a jury in Santa Fe, New Mexico, found that 

 about how it handled third-party acquisition of user data as part of the Cambridge Analytica scandal, as well as over statements it made about misinformation and hate speech. New Mexico is asking for the maximum penalty of $5,000 per violation, which could total over $200 billion.

AI notetaker tool can be treated as a third-party eavesdropper:

, filed in Aug. 2025, is a class-action lawsuit accusing Otter.ai’s AI notetaker tool of violating state privacy laws for joining and recording video calls without users’ consent and using the results to train Otter’s models. On Aug. 13, Judge Eumi Lee 

largely denied Otter’s motion to dismiss

, holding that Otter can be treated as a third-party eavesdropper.

Court blocks the Pentagon’s supply chain risk designation of Anthropic

: On Aug. 27, Judge Rita Lin of the Northern District of California 

held unlawful the Defense Department's designation of Anthropic as a supply chain risk

, finding the action retaliatory under the First Amendment and beyond the department's statutory authority.

Meanwhile, the D.C. Circuit allows the Pentagon to keep its supply chain risk designation of Anthropic: 

On Sept. 25, a divided panel on the D.C. Circuit 

, holding that the agency had “ample support for its conclusion” that Anthropic’s products posed a risk because Anthropic restricts its models from being used for autonomous weapons or mass domestic surveillance.

TikTok settles the federal children’s privacy case: 

announced a $400 million settlement with TikTok and ByteDance

, resolving claims that the company collected personal information from children under 13 without parental consent in violation of COPPA. TikTok was ordered to pay $300 million immediately; the remaining $100 million is contingent on the entry of an order vacating the 2019 consent decree against TikTok's predecessor, 

In the weeks since the $17.1 billion Meta consumer protection settlement, some 

 have argued that the settlement raises serious concerns about—or 

—the First Amendment. They claim that state attorneys general (AGs) engaged in a form of “

”—indirectly seizing control over Meta’s application of its content moderation policies in a way that their congressional counterparts could not have achieved directly through lawmaking.

These critiques resemble a trend that Supreme Court Justice Elena Kagan has 

 “weaponizing the First Amendment,” which expands fringe applications of the speech clause such that the core point of the constitutional protection—fostering individual autonomy and self-governance—is diluted. 

, have pointed out some version of this problem.

Over the past couple of decades, US litigators arguing on behalf of tech companies and industry groups have used the First Amendment as a weapon to strike down consumer protection laws. Consider the 2011 case 

., where the Supreme Court used the First Amendment to strike down a state law preventing drug companies from using data about the prescriptions doctors wrote in marketing designed to get doctors to prescribe different drugs. Sorrell’s innovation was to use the First Amendment to undermine consumer protection.

The novelty here is to apply the playbook to a settlement. But the idea that mutually agreed upon settlements violate the First Amendment through impermissible jawboning risks transforming conduct and products that are not speech (and certainly not speech entitled to heightened scrutiny) into speech for the purposes of the First Amendment. This reclassification is descriptively wrong, and the result is to effectively foreclose consumer protection actions that, once transformed into a speech question, will fail a heightened First Amendment test. This outcome is particularly ironic because the Meta litigation was 

 running afoul of the First Amendment and the third-party content immunity provided by 

Section 230 of the Communications Decency Act

, which have been longstanding roadblocks to addressing the real social harms that platforms cause.

One reason the jawboning argument is descriptively wrong is that it assumes that Meta is an editor whose First Amendment rights might be trampled by this settlement. Maybe Facebook (as the company was known when it was founded over 20 years ago) acted like an editor when it 

 from a reverse chronological timeline to moderating its News Feed, but times have changed.

 (2024), the Supreme Court compared Facebook’s News Feed curation to editorial judgement. Justice Barrett’s concurrence, meanwhile, states that any inquiry required to secure Moody's protections would be “fact intensive.” As such, it’s necessary to consider how much has changed since 2021, when the factual record in Moody was developed. Between then and now, Meta has 

 away from demoting certain categories of “borderline” content (for example, about COVID-19 vaccines) and leaned increasingly on 

 to determine what appears in users’ feeds. This is a practice that the Moody Court 

 as “respond[ing] solely to how users act online—giving them the content they appear to want, without any regard to independent content standards.” And it’s a practice that the Court explicitly “d[id] not deal with.”

Meta’s abdication of editorial functions accelerated during 2025. Meta sunset its 

 in the US in favor of Community Notes, which is driven by users. Meta doesn’t write the notes or “decide which ones show up.” It also limited what it called the “

over enforcement” of its Community Standards

, in part by abandoning its practice of demoting content that might have violated its rules. And, without explanation, it “

” for a host of content types that an editor would weigh in on, including “Content Posted by Repeat Violators of Our Policies, Posts that Indicate Suspicious Virality, and Unsafe Reporting about Suicide.” Thus, the remaining content moderation that Meta appears to engage in seems to deal with illegal or extreme forms of content—a far cry from the “

” that the Moody Court found in the company’s Community Standards and Community Guidelines.

Further, it's unclear the extent to which these judgments are being executed by artificial intelligence, which would further distance Meta from editorial discretion and closer to cold statistical probabilities. In March, Meta 

 that it was shifting to using “AI systems across our apps… transforming our approach," while simultaneously reducing its "reliance on third-party vendors.” While Meta asserts that “we'll still have people who review content,” these changes run directly into the hypothetical that 

 that could undermine the editorial analogy: when “a platform's owners hand the reins to an AI tool and ask it simply to remove 'hateful' content," then does that technological change "attenuate the connection between content-moderation 

… and human beings' constitutionally protected right to '

the ideas and beliefs deserving of expression, consideration, and adherence.’” Indeed, one of the Court’s foundational opinions establishing “editorial control and judgment” as protected by the First Amendment, 

, contemplates “the choice of material to go into a newspaper, and the decisions made as to limitations on the size and content of the paper, and treatment of public issues and public official[s],” all of which are choices made by a person.

Taken together, these moves have shifted the company away from making independent editorial decisions and toward deferring to personalized rankings and probabilistic automation. How fully Meta has abandoned Community Standards for personalized rankings and human judgment for probabilistic automation is a fact-intensive question that courts would need to answer. But the movement toward these practices, both of which de-center humans from the content moderation process, is clear. What Moody has told us is that these are the very practices that the Court excepted from its editorial analogy.

It’s notable that Meta’s policy changes to fact-checking and Community Standards were made in January 2025. If we’re going to talk about jawboning, let’s have a conversation about 

 in the shadow of the Trump administration.

Not only does the jawboning claim fail because it is not clear that the way Meta now does content moderation qualifies for First Amendment protection under Moody, but it also fails in light of the Court’s most recent treatment of this question. In 

National Rifle Association of America v. Vullo

 (2024), the Court held that the NRA plausibly alleged that Maria Vullo, former superintendent of the New York Department of Financial Services, violated the organization’s First Amendment rights when she coerced financial institutions that her department regulated to stop doing business with the NRA and other pro-gun groups to “punish or suppress” their advocacy.

The facts in the Meta settlement are different. Vullo wielded direct regulatory and enforcement power over financial institutions, and she allegedly targeted the NRA’s speech because of its viewpoint. Here, the AGs had a cause of action that moved through the litigation process for several years and Meta—an incredibly well-resourced and sophisticated actor—had the opportunity to contest at every stage. What’s more, the AGs were not focused on Meta’s viewpoint, but rather its product design and data collection practices. (For one articulation of this argument, consider the 

 in Commonwealth v. Meta Platforms, Inc., a Massachusetts case heard earlier this year.)

 to which Meta agreed either continue existing practices, like teen content safety policies, or amount to data and design rules, like the time management provisions. Casting the settlement as violating the First Amendment risks both treating Meta as an editor for constitutional purposes and transforming its product design choices into protected speech acts. Because Meta’s changed moderation practices around personalized rankings and probabilistic automation suggest that it is no longer an editor, or at least it is a far cry from the one it once was, and because framing design as speech forecloses on legitimate products liability and consumer protections theories of accountability, those expansionist arguments should fail.

To be sure, some of the settlement’s provisions might violate the First Amendment. One of the 

 that critics have raised is the restriction of teen users’ ability to see ‘like’ counts on their posts. But because users aren’t a party to the agreement, they aren’t bound by it and can still file separate lawsuits over whatever First Amendment claims they may have against the states. That’s not to say it would be an easy case to make, but the claims are preserved by the settlement.

My point is not to defend all the settlement’s provisions. I share serious reservations about the age verification requirements and how they might consolidate the power of Meta’s surveillance and data collection capacities. Nor do I endorse every provisions’ efficacy. Time will tell.

The point is narrower: it’s worth critiquing the settlement, but we shouldn’t be doing it by reflexively invoking the First Amendment. Personalization and probabilistic automation are not the type of practices that we should treat as editorial judgment or expression. The Moody Court was right to flag those practices as outside of its scope of protection. And for good reason: if we extend speech coverage this way, we risk diluting this freedom by expanding its peripheral applications while eroding its core purposes of autonomy and self-governance. Further, we risk undermining the ability to hold data collection and surveillance companies like Meta accountable at a crucial inflection point where they are transforming themselves into AI companies built on the back of those practices.

Mark Scott is a Contributing Editor at Tech Policy Press.

BRUSSELS — The European Union is having a Jekyll and Hyde moment on tech.

In the corridors of power within the 27-country bloc, two competing philosophies are vying for supremacy just as digital policy has become central to global geopolitics between Europe, the United States and China.

The EU has committed itself to a deregulatory agenda — epitomized by Mario Draghi’s 

 — that aims a bazooka at once-sacred cows like the Artificial Intelligence Act and General Data Protection Regulation to reduce red tape and jumpstart Europe’s sluggish economy.

Yet Brussels also continues to commit itself to new digital legislation, most notably the recent 

, that represent the bloc’s traditional regulate-first philosophy.

Confusingly, both of these opposing forces are playing out, in real time.

, European Commission president Ursula von der Leyen name-checked both the push for deregulation and the new round of digital rules.

First, she boasted that “our 12 omnibus proposals are reducing the administrative burden by around €17 billion a year,” in reference to a series of red-tape-cutting initiatives, including those 

. Then, minutes later in the same speech, von der Leyen announced the EU Kids Act, telling lawmakers: “It is we who decide our rules, not Big Tech.”

For EU officials, this perceived contradiction — in which the bloc wants to cut digital regulation while also proposing new legislation — is not a contradiction. They point to how many of the new proposals, including the EU Kids Act and Digital Fairness Act, are aimed at simplifying oversight, especially for small businesses.

Under this worldview, the new legislation supports Draghi’s simplification agenda, and the review of landmark laws like the EU’s 

 and Digital Services Act is an opportunity to pare back oversight that may become too burdensome.

That may be an argument that holds water inside the 

. But for those outside the European Commission’s headquarters, Brussels’ love-hate relationship with digital regulation can come across as muddled — even confusing — just when European citizens 

In Brussels’ digital policy circles, the European Commission’s logic is under fire from both directions. Consumer and digital rights groups 

 the so-called Digital Omnibus is deregulation dressed as simplification. The tech industry 

 the EU’s simplification agenda into a weapon to attack its new legislative proposals, including framing the EU Kids Act as duplicative regulatory oversight already built into the bloc’s existing rulebook.

Europe’s leaders find themselves caught between two seemingly opposing priorities.

, particularly from US tech giants, many national leaders and EU officials believe Europe must pare back strict oversight of this emerging technology or face being left behind.

 of the world’s AI compute power compared to 75 percent of the global total from the US. 

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe