The Mythos Recall and the Global AI Sovereignty Debate

Good morning!

On late Friday afternoon, Anthropic announced it received a letter from the US Department of Commerce notifying the company that the government had “issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.” To satisfy the government’s demand, the company said it had disabled access to the two models for all its customers. Yesterday, I published an early analysis of the situation, including on reaction from foreign capitals, where fears over AI sovereignty and reliance on American AI were already roiling. 

As the G7 summit kicks off in Évian this week, AI sovereignty is emerging as a key fault line between Western allies. Before the Anthropic news, Mark Scott, Tech Policy Press contributing editor, wrote about how the US, EU and Canada have outlined competing visions—and why allies increasingly want to reduce their dependence on Silicon Valley. “The likes of the EU and Canada are willing to invest billions of dollars in public money to support domestic data centers, semiconductor manufacturing and local alternatives to Anthropic’s Claude or Google’s Gemini services,” he writes.

The Anthropic situation also exposes the lack of comprehensive AI regulation in the US. Two members of Congress are trying to address that. On June 4, Reps. Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.) released a 269-page bipartisan discussion draft of a bill called the Great American Artificial Intelligence Act of 2026. I invited them to the podcast, and was able to speak with Rep. Trahan last week about the bill and about some of the criticisms of it, including a provision on preemption of state laws that would regulate frontier model development. Listen to the podcast here.

That's not all this week. Read on for more.

Surveillance, spyware and privacy

• From Gaza to Iran, documented cases show spyware increasingly being used for battlefield intelligence, military espionage, and influence operations, but researchers warn the public record may capture only a small part of its use, writes Tech Policy Press fellow Vas Panagiotopoulos.
• The UK health system is rolling out a Palantir-built data platform that consolidates fragmented patient data, while experts raise concerns about transparency, oversight, and potential use in future, report Jade-Ruyu Yan, Tech Policy Press UK fellow, and Aman Sethi, editor in chief of openDemocracy. “In principle, all that’s stopping the Home Office from accessing NHS data are legal safeguards that can be reversed,” they write.
• Congress let FISA 702 lapse on June 13 rather than allow votes on reform — exactly the false binary that Tech Policy Press fellow and Center for Democracy & Technology Security and Surveillance Project deputy director Jake Laperruque had warned against. He argued lawmakers should reject the all-or-nothing framing, allow votes on bipartisan amendments, and pass meaningful safeguards.
• Across countries and platforms, AI’s outsourcing maze depends on data workers under surveillance-heavy conditions, with accountability fragmented across multiple firms, reports Marché Arends. “Contracts can be pulled at a moment’s notice with no avenue for recourse, leaving workers adrift. These are workers who are essential to AI’s development but are completely stripped of their power in the process,” she writes.
• A blurred face isn't hidden—it's degraded. But AI trained to recover a signal from noise is good at restoring what's been degraded. shirin anlen of WITNESS and Gabi Ivens of Human Rights Watch explain why human rights groups must trade the cosmetic blur for structural anonymization to protect people.

AI, chatbots, the information environment, and more

• As the 2026 midterms approach, how are the biggest AI firms preparing? Tim Bernard, a tech policy analyst and writer, breaks down the election-safety plans from OpenAI, Google and Anthropic—from voting-info partnerships to watermarking to cyber defense—and where the gaps still are.
• Google calls it the biggest Search overhaul in 25+ years. AI Mode now has 1B+ monthly users—many defaulted in. Elise Silva, director of policy research at the University of Pittsburgh's Institute for Cyber Law, Policy, and Security, asks what this shift means not just for competitors, but for billions of users, our shared information environment, and democracy.
• The UK's CMA just forced Google to let news publishers opt out of AI Overviews without vanishing from Search. Martha Dark, co-executive director of tech justice nonprofit Foxglove, argues this is a landmark win — and a blueprint regulators in the EU, Brazil and beyond should follow next. “When a company like Google — with more money, power and influence than many nation states —can impose terms that publishers cannot realistically reject, regulators have a duty to intervene,” she writes.
• Protesters with "AI don't vote" signs. Vendors hawking chatbots. Activists fighting data centers. Tech Policy Press contributing editor Dean Jackson considers how progressives are wrestling with AI in his report from Netroots Nation in Philadelphia
• Millions are using chatbots regularly—including for mental health help. But the public, clinicians, and policymakers still know little about the risks. Chris Mills Rodrigo, Tech Policy Press fellow, on the dangerous knowledge gap and what's being done to close it.
• Dr. Caroline Figueroa, a Public Voices Fellow on Youth Well-Being and Power with The OpEd Project and Hopelab, writes that when it comes to teens and chatbots policymakers are focused on acute harms, but the bigger concern may be emotional dependency on AI itself— and ensuring teens receive the human support they need. “Policymakers could further mandate social and emotional AI literacy education so young people understand which advice not to trust and the limits of AI as a source for personal support,” she writes.
• As lawsuits over AI chatbot-related harms mount, lawmakers are seeking to protect minors online. States are leading with new safeguards while efforts in Congress to advance federal rules remain slow going, with competing measures under debate, report Erika Tulfoand Yuqing Liu of the Medill News Service.
• “Young people are resilient, but too often, the kids are not alright.” AFT President Randi Weingarten's call to limit tech in classrooms signals a growing urgency. Georgetown University’s Meg Leta Jones and Megan Beam unpack the union's 10-point plan and the wider fight over AI and screens in schools.
• The UK's push to ban social media for teens is gathering pace, but critics say key questions about privacy, enforcement, and effectiveness remain unanswered, reports Tech Policy Press UK fellow Jade-Ruyu Yan.
• Young people in India are discovering their politics — and much of it emerges through, and is directed at, digital technologies. Apar Gupta, Tech Policy Press fellow and founder director of the Internet Freedom Foundation (IFF), on how a test data scandal became a youth reckoning.
• AI governance keeps collapsing two different failures into one: imperfect data and imperfect systems. They are not the same problem, and they demand different forms of accountability, reform, and oversight, argues Michael A. Santoro, professor of management and entrepreneurship at Santa Clara University.
• US export controls were meant to preserve the country’s global lead in AI, writes Innovation Technology and Innovation Foundation’s Daniel Castro. Instead, he argues they may be accelerating China’s push for technological self-sufficiency and strengthening competing AI ecosystems. “America's objective should be clear: maintain leadership in AI, preserve leadership in semiconductors, and strengthen economic security,” he writes.

Regulation and online safety

• Two new Brazilian executive decrees tackle online fraud, gender-based violence and AI-generated intimate imagery. Nina Santos, Deputy Secretary for Digital Policies, and João Brant, Secretary for Digital Policies at the Brazilian Presidency, on why it matters in Brazil and beyond.
• Ireland’s upcoming EU presidency highlights tensions between simplifying digital regulation and expanding protections on AI, data use, and child online safety, writes Tech Policy Press fellow Liz Carolan.
• Check out another installment of the Global Digital Policy Roundup for May 2026 from the experts at Digital Policy Alert. Maria Buza and Tommaso Giardini highlight tech policy developments in content moderation, artificial intelligence, competition, and data governance.
• Appeals Centre Europe’s Thomas Hughes writes that his organization's latest Transparency Report offers a glimpse of Article 21 of the Digital Services Act’s potential — and the work still needed to ensure independent review of platform decisions.

Finally, a special note of thanks to Cristiano Lima-Strong for all of his contributions at Tech Policy Press. He is joining Bloomberg to cover financial services. We wish him the best!
Cheers from New York, where people are still glowing after last night's Knicks win. I'm sending that energy your way!

-Justin