This week's Meta settlement will reverberate across the world for years

Good morning!

Some weeks it’s difficult to prioritize what news to focus on in the first paragraphs of this newsletter. This is not one of those weeks. Wednesday’s announcement of Meta’s landmark settlement with attorneys general from US states and territories over teen safety on Instagram and Facebook will reverberate for months, if not years.

As Tech Policy Press contributors and readers digested the terms of the settlement and its implications, many expressed skepticism.

“There are provisions here that could make a difference for children like the time limits, nighttime restrictions, age assurance, changes to certain engagement features, and auditing,” J.B. Branch, the director of federal AI governance and technology policy at Public Citizen, told me in an email. “But Meta has an extraordinarily long record of apologizing after the damage is done, paying fines or settlements, and moving on. So I am not prepared to take promises of transformation at face value.”

We covered the settlement in multiple ways. Tech Policy Press fellow Varsha Bansal and I took a close look at its terms, including the “Phase II” concept of “industry wide adoption” that would extend limitations on child usage to other industry players if they agree to enter into similar obligations. And the day after the settlement, we hosted a live recording of the Tech Policy Press podcast featuring a discussion with the Knight-Georgetown Institute's Alissa Cooper, Tech Justice Law's Meetali Jain, and Emory Law's Matthew Lawrence.

One thing we heard from the participants in the podcast and others who spoke to Tech Policy Press this week, including Arturo Béjar, a former Meta child safety leader who testified at the trial, is real concern over the provisions in the settlement for an independent auditor. Many fear that the mechanism, as described, will fail to deliver meaningful public disclosures and that we will not be any closer to understanding the actual harms from Meta’s products. That matters for many reasons, including that it may limit how well limitations are calibrated against the need to protect free expression. 

Tech Policy Press deputy editor Ramsha Jahangir considered the view from abroad, reporting on what the Meta settlement means for Europe's approach to child online safety. “This settlement has implications far beyond the US,” Owen Bennett, an independent tech policy expert in the UK, told Ramsha. “If I were a regulator sitting in Brasilia, Brussels, or Canberra, I would see this as a huge opportunity,” Bennett said. The settlement could have significant impact where bans are being considered (a point considered Friday in The New York Times, which referenced our global social media age restriction tracker.)

And what does it mean more generally for the rest of the world? In a perspective piece written in her personal capacity, Nighat Dad, founder and executive director of the Digital Rights Foundation and a member of the Oversight Board, argues the rollout of the protections should follow the harm, which is global. The big social media platforms “are growing in Jakarta, Lagos, São Paulo, Dhaka and Karachi, not Ohio,” she says. “A standard limited to the US asks them to change where they have the least to lose. A global standard puts pressure where their businesses are growing and lets Meta set the rules instead of simply following one US agreement,” she writes.

Finally, we published a perspective on the settlement-within-a-settlement that you might have missed unless you read the entire document. The word "Cambridge" appears 45 times in the document, notes Jason Kint, the CEO of DCN. Buried inside it: a $459,293,017.80 payment resolving Cambridge Analytica claims across at least 46 states and two territories. “The fact that Meta is still paying to resolve Cambridge Analytica claims in 2026 shows that the scandal was never really just about Cambridge Analytica. The scandal raised fundamental questions about Facebook,” he writes.

Meta is not out of legal jeopardy over child safety issues (nor on Cambridge Analytica-related matters), and cases on these issues will continue to advance. But the chess board looks significantly different than it did at the beginning of last week. We’ll have more on the settlement and its implications in the days and weeks to come.

More on age bans and children’s online safety

• France's Constitutional Council struck down the country's under-15 social media ban on Aug. 14. Christian Cirhigiri of CDT Europe and Sabine Witting of Tech Legality unpack the ruling and what it means for Europe's approach to social media bans. “Looking at Brussels, the Constitutional Council’s decision should force a much-need reflection moment on the necessity for the European Commission to grapple with the extent of alignment between its anticipated EU-wide framework and the protection of fundamental human rights of minors and children as well as of all users,” they write.
• As California's Legislature moves toward a four-year ban on AI toys, Sameer Hinduja, a professor of criminology and criminal justice at Florida Atlantic University, argues that a moratorium buys time without producing the evidence or the standards that would actually protect children.
• Brazil's data protection authority fined TikTok $29.7 million this week for failing to protect children's data, the first time the ANPD has sanctioned a major tech company. Tech Policy Press fellow Tatiana Dias on what the case reveals about Brazil's expanding regulatory powers.

AI accountability and the public interest

• Amid the ongoing debate over how to judge the safety of frontier models, Axiomatic AI CEO Jake Taylor, who helped establish the US Center for AI Standards and Innovation at the National Institute of Standards and Technology, argues US AI policy needs a key pillar: powerful systems require consistent verification against safety benchmarks that are public and interpretable. We need “evaluation teams working inside and alongside the frontier labs,” says Taylor.
• One thing is clear—voluntary blog posts and self-audits aren't accountability, argues Darryl Slabe of ERA Cambridge. He lays out a purpose-built corporate offense for AI harm, modeled on the UK's failure-to-prevent-fraud law, that shifts the burden onto the lab. Following disclosures that systems developed by OpenAI, Anthropic and Meta intruded into third-party services, the question is how they could be brought to justice. “Whether OpenAI’s, Anthropic’s, or Meta’s precautions were reasonable in these cases should be a question for a criminal court,” says Slabe.

Addressing structural problems in AI governance

• Kevin Frazier, director of the AI Innovation and Law Program at the University of Texas School of Law and a senior editor at Lawfare, argues that uneven AI access is bifurcating the US "into those that use AI and those who bear its costs," and Congress must respond. "It's a public imperative that such a disparity does not become entrenched," he says.
• Daniel Arias Rivera, a Colombian lawyer and human rights specialist, explains how Bill 025 copies the EU AI Act's risk-based architecture without the institutional capacity, technical expertise, or enforcement mechanisms to make it work in Colombia.
• After a massive exam scandal triggered mass protests, India's government responded with a plan to make its examination system more technology-driven. Tech Policy Press contributing editor Amber Sinha on why that response misses the deeper crisis.

Other must-read pieces this week

• With US debt past $40 trillion and Treasury interventions failing, Cole Donovan of the Stand Up for Science Foundation argues fiscal consolidation is coming — and experts keen to rejuvenate science and technology policy need to pay attention. “The blowback against data centers should serve as a warning of what will happen if policymakers impose financial and physical costs on society without earning the public’s trust,” he writes.
• "Data center discontent" is now a statehouse force. Tech policy analyst and writer Tim Bernard tracked nearly 400 bills in 12 months as lawmakers stopped courting developers and started demanding decommissioning bonds, NDA bans, and moratoria.
• There is still more that unites Brussels and Washington on tech than divides them. The transatlantic trust deficit is too large and too politically charged for either side to admit it publicly, writes Tech Policy Press contributing editor Mark Scott.

Announcements

• Applications for the 2027 Tech Policy Press Fellowship are now open! The year-long, part-time fellowship is designed to support mid-career journalists, researchers, and public policy professionals pursuing independent reporting and analysis that explore urgent questions at the intersection of technology and democracy. Selected fellows will have the opportunity to join a global cohort of experts from diverse fields, collaborating across disciplines to share insights and help inform the most pressing debates in tech policy. The deadline to apply is September 24, 2026.
• ICYMI: We’re extremely excited to share that Solana Larsen has joined Tech Policy Press as managing editor! Solana has more than two decades of experience working in senior editorial roles at publications and with communities dedicated to analyzing technology’s role and impact on society, including with Mozilla, Global Voices, openDemocracy, the Open Knowledge Foundation, and more.

I wish you the best for the week ahead!

-Justin