An array of colorful, fossil-like data imprints representing the static nature of AI models, laden with outdated contexts and biases.

Luke Conroy and Anne Fehres & AI4Media / 

A Case Against Forever: Why Cloud and Social Media Need Inactivity-Based Deletions

Howard Lutnick, US Secretary of Commerce, testifies before the House Committee on Appropriations Subcommittee on Commerce, Justice, Science, and Related Agencies in the Rayburn House Office Building in Washington, DC on Thursday, June 5, 2025. (Photo by Aaron Schwartz/Sipa USA)

Did the US Government Just Set An AI Export Precedent by Blocking Mythos?

An abstract image based on the US Constitution

Earning a Social License for Transformative AI

BRUSSELS, BELGIUM—JUNE 3: EU Commissioner for Tech Sovereignty, Security and Democracy Henna Virkkunen introduces the European Technological Sovereignty Package, featuring the Chips Act 2.0, Cloud and AI Development Act, and related strategies to reduce the EU's dependence on non-European suppliers for critical digital technologies. (Photo by Thierry Monasse/Getty Images)

Europe’s AI Sovereignty Problem Runs Far Deeper Than Frontier Access

Rep. Lori Trahan (D-Mass.) speaks during an interview with POLITICO in her office on Capitol Hill on May 21, 2026. (Francis Chung/POLITICO via AP Images)

Unpacking the Great American Artificial Intelligence Act of 2026

Anthropic's Co-Founder and CEO, Dario Amodei on stage at the Code with Claude developer conference on Wednesday, May 6, 2026 in San Francisco. (Don Feria/AP Content Services for Anthropic)

Anthropic's Mythos Recall and the White House's Missing AI Safety Playbook

G7 Summit Set to Kick Off Amidst Allies' Widening Rift Over AI Sovereignty

A collage featuring a vintage illustration of a woman’s head mapped with labeled sections resembling a phrenology chart. The mapped sections are overlaid by a neutral network diagram– depicting crisscrossing black lines. Two anonymous hands extend from the left side, pulling on two wires from the diagram.  In the background is a panel of the Turing Machine with numerous knobs and switches, highlighting a connection between the history of computing, psychology, biology, and artificial intelligence. 

Personifying AI Harms People and Protects Companies

Helen McEntee, Ireland's Minister for Foreign Affairs and Trade, outlines priorities for Ireland's upcoming EU Council Presidency on June 10. Source: Irish Presidency of the Council of the EU 2026.

Ireland Targets AI Omnibus, Child Safety in EU Presidency Agenda

Northwest view up to the pediment, rotunda, and dome of the California State Capitol in Sacramento. (Radomianin / Wikimedia Commons)

Northwest view up to the pediment, rotunda, and dome of the California State Capitol in Sacramento. 

 States Step in Where Congress Stalls on AI Safeguards for Kids

A man showing mental distress from constant exposure to harmful content online. His family, in the background, progressively disappears.

How AI’s Labor Supply Chains Fail Workers

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

Global Digital Policy Roundup: May 2026

Abhijeet Dipke, center, founder of the Cockroach Janta Party addresses supporters during a protest demonstration in New Delhi, India, Saturday, June 6, 2026. (AP Photo/Manish Swarup)

After Test Data Scandal, India’s Youth Stake Their Claim on the Digital Square

A blur effect added to Leonardo da Vinci’s Mona Lisa

AI Can Rebuild Blurred Faces, So How Do We Protect People Now?

Close-up of Keir Starmer, wearing a dark suit, white shirt, and tie, looking directly at the camera.

Prime Minister Sir Keir Starmer delivers opening remarks ahead of a meeting with senior figures from TikTok, X, Meta and other social media giants at 10 Downing Street, Westminster. Picture date: Thursday April 16, 2026. 84281335 (Press Association via AP Images)

Could a UK Teen Social Media Ban Work Without Expanding Surveillance?

The image shows a hand holding a smartphone in front of a computer screen with war and other words projected across.  

Hidden Spyware Use in War May Be More Widespread Than We Realize

How OpenAI, Google, and Anthropic Plan to Handle the 2026 US Midterms

Panelists at a session titled “Does AI Change How Change Happens?” at Netroots Nation in Philadelphia on June 6, 2026. Dean Jackson/Tech Policy Press.

At Netroots Nation, Progressives Divided on AI

LONDON, UK—NOVEMBER 15, 2025: Panorama of Google logo sign on door of HQ building in Kings Cross.

The UK Just Said No to Google's AI News Grab. Who's Next?

Three colorful GPUs with their packaging cleanly removed laying on a white surface.

The China Chip Strategy That Is Backfiring on America

 is published in partnership with the AI Now Institute.

This weekend, the United States administration 

 to cut off foreign access to two of its advanced AI models. In Europe, the news has been interpreted almost unanimously: a wake-up call that shows dependency means access can be cut off overnight.

, we make the case that Europe’s AI market is 

 with the ecosystems of dominant US players, in ways that boosting both supply and demand alone cannot disentangle. As a result, interventions that seek to secure sovereignty in AI risk leading to much more entrenched dependence. The first part of this series looks at the demand-side of things.

The limits of AI sovereignty at the application layer

 is lauded as one of Europe's biggest AI success stories — and understandably so.

Founded in Stockholm in 2023, it was quickly 

 as the fastest-growing software startup in history. Currently, the company is 

 at $6.6 billion after raising a total of $540 million and is frequently cited alongside Mistral as proof that Europe can build globally competitive AI companies. Lovable lets anyone build a website or an app (“something they love”) by chatting with an AI, and lots of people do.

But Lovable’s rise is dependent on the kindness of strangers. Instead of developing their own models, deep technological moats, or exploiting a local technology stack, the company builds on the American tech stack: prompts are routed to models made by Anthropic and Google and distributed in the existing hyperscaler ecosystems. For example, on June 4, Lovable 

 a deepened partnership with Google Cloud that is supposed to lead to a “fivefold increase” in Lovable's Google Cloud footprint, with recent 

 of the products into the Google Workspace.

Let us be clear. There’s nothing wrong with focusing on making AI useful, and building on top of existing models. It’s also hard, requires speed of execution, great UX, nurturing a community of users, and getting there first. As a long-term business strategy, however, it carries a risk.

 that Anthropic created a new in-chat app builder that lets users generate applications from simple prompts — essentially competing with their client Lovable’s entire product. Weeks later, Anthropic launched 

, a law-focused offering that put legal AI startups on the defensive — including Harvey, Legora, and Robin AI, all of which deploy Claude models at the core of their products. The supplier had become a competitor twice over within a matter of weeks. This pattern — an incumbent absorbing the feature a smaller company built on top of it — is known as getting “

,” is a well-known, yet often-forgotten trope in the tech industry, named after Apple introduced a new feature that rendered a popular third-party tool irrelevant. 

Not all European start-ups build on proprietary US foundation models: some train their own, some run smaller, domain-specific models on proprietary data and on-premise infrastructure, or focus on physical-AI. However, we begin this series by interrogating application-layer companies that primarily create end-user software and workflows on top of existing foundation models, rather than training frontier models or operating large-scale AI infrastructure. That is because the application layer is a specific bet European policy has chosen to place.

Lovable is an example of a broader trend in European tech strategy. It is based on an assumption that Europe's opportunity in AI lies in building tools that make AI useful at the application layer (rather than spending billions on building deep technology, compute infrastructure and parallel ecosystems). The application layer is where most of the profit will be made, so the argument goes, and focusing on applied AI helps Europe stay in the ‘AI race’ by essentially piggybacking on the investments of others (

 by SAP’s CEO in the Economist and these 

 by Skype founder-turned-venture capitalist are prime examples for this position).

This idea is genuinely tempting. The bet (call it “sovereignty-at-the-application-layer”) provides a profitable way forward without playing the prohibitively expensive game of competing at the frontier-model layer. Asset-light software companies, with low headcount and few hard, illiquid assets, are a far better fit for how venture capital is built to allocate money. It also pattern-matches to how the internet era seemed to play out. From travel platforms like Booking.com to streaming services like Netflix, major winners of the internet economy emerged at the application layer.

The idea is also alluring to policy-makers. “Applied AI” first entered the European policy world with von der Leyen’s 

Political Guidelines for the 2024-2029 Commission

, which set out the intention to build an Apply AI Strategy to boost new industrial uses of AI.

But the policy argument conflates financial efficiency with strategic autonomy. There are three serious arguments that speak against the long-term sustainability of the application-layer bet for Europe: creeping supplier competition, loss of control and value capture upstream

Suppliers as competitors and ultimate beneficiaries

Anthropic competing with their client, Lovable, isn’t an isolated incident. There’s been an industry-wide trend of model-makers entering the application market, driven by the need to capture value beyond the raw model provision. Claude Managed Agents (alongside Claude Cowork) are a direct answer to the self-hosted infrastructure for autonomous agents developed by OpenClaw (the free and open-source AI agent that can run on device). OpenAI tried to acquire coding assistant Windsurf (which was ultimately 

Not just model makers, but also dominant hyperscalers are both suppliers and potential competitors for application-level start-ups. Amazon, Google, and Microsoft all compete at the application layer directly, Microsoft on office and productivity software integrated into Office 365, Google across search and productivity, and AWS now also on industrial applications of AI. All three companies also sell AI as a service bundled into their cloud platforms, making it excruciatingly difficult to compete with these players.

This is what distinguishes AI from earlier software eras. Companies like Salesforce and Adobe could build lasting application-layer franchises because their suppliers sold commodity inputs and had no designs on their markets. AI applications rent their defining capability from a handful of firms that are expanding into applications themselves, and that set the marginal cost of every customer interaction.

If all of this sounds familiar to astute historians of Europe’s relationship with dominant tech companies, that’s because it is. Europe already has its application-layer success story, and it is a tenant. Spotify, its most celebrated consumer-tech champion, runs on someone else's cloud, hands most of its revenue to the rights holders it cannot do without, and has spent years fighting Apple over App Store fees, payment restrictions, and market competition. Succeeding at the application layer and achieving independence turn out to be different things.

Application-layer AI companies inherit an even harder version of this. Their marginal costs are rigid and set upstream by model providers. And while they could theoretically switch to open models, once these reach parity with proprietary ones, leaving is hard, often by design, and the financial and infrastructural dependencies that are being built are hard to disentangle. As a result, these companies bear far more resemblance with a different corner of the internet economy: the flight-comparison sites that ran on Google traffic until Google Flights displaced them; the publishers who reorganized around Facebook's algorithm and pivoted to video, until 

 the engagement figures had been vastly inflated by Facebook; and apps like Spotify that remain subject to Apple’s ever changing App Store terms, 

and a cut of up to 30% on many transactions

Today, we’re seeing this exact dynamic play out at a much larger scale. European AI companies building on American APIs risk becoming something like an outsourced R&D department for leading model providers: they take the risks of discovering which applications actually work, while the value of those discoveries flows back down the stack. The moment a European startup comes up with a seriously profitable idea, the model makers have every incentive to acquire the company or simply rebuild the product as a feature.

This is bad for Europe’s long-term independence, but for the typical venture capital cycle, selling early to a wealthy incumbent is an attractive way to reach exit liquidity — and resisting that pull is hard for a startup stacked with VC. Lovable currently has 54, mostly venture investors in its cap table, with Google’s growth venture arm CapitalG leading the latest funding round. Instead of battling for market share against stupefyingly wealthy incumbents, selling to these companies provides a successful exit for the early investors and a small team of founders. This points to the 

 between the strategic objective of technological sovereignty and the 

 of venture capital financing. While the European AI ecosystem has notable 

, this typical VC model still skews startups’ trajectory towards the US ecosystem.

It is often noted that European startup founders invest in a new generation of founders, 

. But this flywheel might never spin free of the structure of which it is part. It might lead to a cycle of serial exits to the balance sheets of US tech ecosystems, without generating enough escape velocity to create alternative gravity wells that could serve as a scalable platform anchoring European companies to pursue pathways. Against this gloomy picture, a counterargument is routinely offered: European companies can protect themselves through proprietary data, scale up fast enough to become dominant themselves, expand through mergers, or form a captured user base into long-term contracts. Moreover, the irreplaceability of context-specific data generated by proprietary know-how and workflows of vertical industries and incumbents 

 the insulating layer protecting the European companies from US competition, and creating staying power in the European ecosystem

But these mythical, vertical-specific data lakes in manufacturing or healthcare are not available for horizontal startups such as Lovable, which have to depend on customer traction, user experience, and speed as their differentiating factors. These are relatively weak moats compared to the financial dominance, ecosystem control, and integration advantages of existing platforms that control the engine on which these corporations depend. The first-mover advantages are real, as OpenAI testifies. However, the first to move is not necessarily the last to survive.

Infrastructure and model layers competing with application-layer start-ups that depend on them isn’t the only reason why focusing on the application layer is risky. When delegating a core part of their infrastructure, companies also delegate control over pricing models, and APIs can change overnight, while model behavior can be altered without notice. This leaves a fundamental vulnerability at the heart of an API-based business model. Recently, for instance, Anthropic 

 from a flat rate to usage-based pricing, destabilizing application builders downstream.

stake their hopes on staying “provider-agnostic”

: switching between models rather than tying their products to one. The bet is intuitive, but it leaves them exposed on two fronts at once. On prices, API costs are currently depressed sector-wide as providers still fight for market shares. Anyone building on top of these APIs is heavily exposed once prices reflect real costs, or when the model layer 

 starts to capture value and is able to extract rent. On performance, reaching the absolute competitive frontier for a given use case may require model-specific optimization. An agnostic start-up has to forgo these optimizations to prevent lock-in

A counter-argument is that orchestration layers that sit between the app and several model APIs could make switching between models more seamless — and many start-ups, such as the German 

, now offer exactly this. But that raises the question of who sets orchestration standards, and AI companies at the application layer would still be dependent on multiple, though now competing, oligopoly of model makers.

The most serious counterargument is open-weight model parity. Simply put, if open-weight models reach a good enough quality, on par with leading, proprietary frontier models, then application-level AI start-ups can self-host competitive models and build on top. But the largest open-weight models are still released by a small number of AI labs, which means dependence on these models being released under truly open licenses. Most open-weight models 

 in the sense of open-source, and even maximally open models do not by themselves guarantee competition, as the inputs needed to train and deploy large-scale AI remain concentrated in the same handful of corporations, who have a history of using the rhetoric of openness to entrench, rather than dilute their dominance. While open-weight models do shape the bargaining position, the ecosystem is an integrated whole – the distribution layer can shape model and infrastructure choices of application companies squeezed in between.

How these technical developments and advances will unfold is still uncertain. But crucially, we caution against betting that any of these technical changes will automatically solve the underlying political economy that takes into account the broader ecosystem in which those models live.

From an industrial policy perspective, the question of where value flows is perhaps the most important one, and the one that’s often sidelined. 

The Commission's report supporting the Apply AI strategy

, for instance, counts over 4,100 AI startups in the EU with a combined enterprise value of €161 billion. But it counts startups, not supply chains and tracks where AI companies are headquartered, not where value is created or captured.

The ambition is to enable the next Google, or at least the next OpenAI. But Google's power was never the product at the top of the stack, it was owning almost everything beneath it, from chips to cloud to models to data. And the leading proprietary models are each tied to one of the dominant hyperscalers, for instance, OpenAI with 

. As a result, any start-up that builds on leading, proprietary models incurs not just ongoing costs for tokens, but also for inference and infrastructure. For application-layer companies, those costs consume a substantial share of revenue, which flows back to the US model providers and cloud platforms. The access to distribution might also come with strings attached – for instance, Lovable's Google Cloud deal comes with expanded access to Google's own Gemini models and Anthropic's Claude models, where Google (via its investment arm Capital G) 

. To support their expansion of market share, they are motivated to direct token demand to these foundation models by using their investment partnership to determine model choices. As a result, control of margin sits upstream, not at the application layer, vulnerable to being squeezed at will by their core technology provider.

What this means for EU AI industrial policy

This dynamic changes a few core assumptions that are often implicit in EU AI industrial policy: Simply increasing the number of AI start-ups in Europe will not necessarily lead to more independence from US tech. European AI start-ups at high valuations don’t necessarily mean that Europe is building a sustainable AI sector that is here to stay. A thriving scene of European AI start-ups doesn’t automatically mean an economic upside for Europe. And more private investment, especially money bound by the logic of venture capital, does not automatically produce gains that stay in Europe.

The problem is not that European founders build on the best available models. The problem, for the long-term sustainability of Europe’s economy, is that countless rational and perfectly reasonable individual choices further entrench Europe’s dependency in ways that may be even harder to unwind than the status quo. The seeds of European dependency lie not only in the coercive force of the monopolies of the digital economy. They are also sown as a byproduct of ideas that appear self-evident enough to be left unexamined.

When Europe accepts that European companies should build on market-leading foreign partners for fear of being left behind; when hyperscaling venture capital is elevated as the sole viable financial driver of European tech; and when the spirit of Silicon Valley startup handbooks — built in California, by California, and for California — is carried as the absolute model to emulate, the path is largely set. 

The question is not whether Europe can and should build applications on top of existing models — it is rather to note that applications alone do not solve the question of technological sovereignty that Europe so desperately seeks.

, the Commerce Department’s Bureau of Industry and Security issued an export control directive to Anthropic to suspend access to its Fable 5 and Mythos 5 models for any foreign national, whether located abroad or inside the United States, citing national security concerns. Because the company could not reliably screen users by nationality, it 

In the immediate coverage of these events, so far, most of the attention has been paid to questions about the models, the reported jailbreak that prompted the action, the White House’s deliberations, and the company’s response. What’s needed now is policy analysis. And from that perspective, it’s clear that the novel development here isn’t another intrigue about a Anthropic and its large language models. It’s the legal instrument the White House deployed to force Anthropic’s hand, and the precedent it sets going forward.

Export controls have long reached more than physical goods. Software, source code, and electronic transmissions of non-public technical data all fall within the 

, and the policy debates post-encryption suits of the 1990s settled that intangible 

. So the surprise in this scenario isn’t that software can be export-controlled at all. It’s that these authorities, built around discrete transfers of an identifiable item from one party to another, are being applied to a continuously available frontier model reached by API. When the controlled thing is a service that anyone can call from anywhere at any time, the familiar questions get harder. What exactly is the export? The model weights, the act of inference, the capability itself? The regulations have answers for releasing a file or transmitting a dataset. A model that responds to a prompt in real time fits those categories awkwardly, and that friction is now being worked out through enforcement rather than rulemaking.

That's worth dwelling on, because there's a real difference between AI policy that takes shape through a public regulatory process, with notice and comment and the chance for affected parties to weigh in, and AI policy that takes shape through ad hoc actions over time. With a technology this new and moving this fast, the second path is often where things start, simply because events outrun the rulemaking process. But it has a quiet consequence: each enforcement decision becomes precedent, the accumulated precedents begin to function like a rule, and a framework can end up assembled before anyone has stepped back to define it as one. That's not unique to this case or this administration. It's a recurring feature of governing at the frontier, and it's worth recognizing for what it is.

Much remains unclear about this specific situation. Almost everything in the public record is secondhand, much of it from parties with something riding on how this is understood. The directive arrived by letter, which has not been made public, so the precise scope and rationale are known mostly secondhand, as well. The trigger itself is contested. The government acted after another company reported jailbreaking Mythos in a way that alarmed officials. Anthropic characterizes the episode as a 

and says the behavior at issue is widely available from other deployed models, and at least 

one security researcher who saw the underlying work

 disputes the "jailbreak" framing altogether, describing it instead as defensive research. Further, reporting has since suggested the controls were prompted partly by 

concern that a China-linked group had accessed Mythos

, with the worry being that an adversary could reverse-engineer or distill the model, though that account is single-sourced and Anthropic says the issue was never raised. That factual uncertainty isn't a side issue. It's central, because when the determination is made privately and the facts are this contested, it becomes genuinely hard to tell which theory of control the government is even applying.

And underneath all of it sits the question that actually determines how broad this gets: how does a government decide when a frontier model crosses the line into something that warrants control? There are two basic approaches, and they lead to very different worlds.

The first is an incremental-risk approach, where the question is whether a given model materially expands what an adversary could do beyond what's already available to them elsewhere. If similar capability is already widespread, controlling one more instance of it accomplishes little, so this approach tends to focus attention on genuinely novel capabilities and exceptional cases, and it keeps controls relatively rare.

The second is a capability-based approach, where the mere presence of a sensitive capability is the trigger, regardless of whether comparable capability already exists in the wild. This one is potentially far broader, because nearly every frontier model possesses some capability that could be characterized as sensitive, and once you adopt that framing the universe of models potentially subject to control expands dramatically.

Strip away the mechanics and a real question is whether the government is regulating net-new capabilities or regulating capabilities period, and that's the distinction that will determine how expansive this approach ultimately becomes. The current case sits right on that fault line. Anthropic's defense is essentially an 

: whatever Mythos can do, comparable models can already do, so controlling it denies little.This view seems to inform the perspective of a group of cybersecurity experts who 

 urging the government to reverse its position. A capability-based view would find that beside the point, because the sensitive capability is present regardless of what else exists.

Both have a serious argument behind them. The case for incremental risk is that controlling capability already available everywhere imposes real costs on US firms and their allies while doing little to deny anything to a determined adversary. The case for the capability-based view is harder to dismiss than its critics admit, because you can't un-export a capability once it has proliferated. Waiting for proof that a model is truly net-new may mean acting too late to matter. If the goal is preventing irreversible harm, a government may reasonably decide it can't afford to wait for certainty.

That tension doesn't resolve neatly, and pretending otherwise would oversimplify things. But the choice between the two framings is the actual policy decision, and right now it's being made implicitly, buried inside discrete enforcement actions that read like one-offs. This isn't for lack of institutional machinery. The Center for AI Standards and Innovation already runs pre-deployment evaluations of frontier models under agreements with the major labs, and has examined what those models can do with their safeguards stripped away. The capacity to assess capability exists. What's missing is a published standard linking those assessments to consequences: which findings trigger which restrictions.

Which raises a further problem policymakers haven't fully grappled with: do the agencies making these calls even have the technical capacity to make incremental-risk determinations at the frontier? Judging whether a model materially expands an adversary's options requires deep, current knowledge of what already exists across the global landscape, and that's a demanding analytical task. A capability-based approach is, among other things, far easier to administer, which creates a quiet institutional gravity pulling toward the broader framework whether or not it's the right one.

None of this is an argument that frontier models should never be controlled. Some capabilities will warrant serious restriction, and governments have a legitimate role in drawing those lines. The point is narrower: the question of which theory of control applies is a big one, and it's worth resolving deliberately rather than letting it settle case by case. Export authorities built for physical goods and an earlier kind of technology are being asked to do something genuinely new, and that's uncharted ground for any administration.

So the real question isn't what happens to one model. It's which theory of control takes hold, and whether it is set deliberately or emerges over time from the accumulation of individual actions. Much of the necessary infrastructure is already in place. The remaining step is a deliberate one: to define the standard that ties it together before the answer settles by default.

Justin Hendrix is editor of Tech Policy Press. This piece is co-published with 

Anthropic's Co-Founder and CEO, Dario Amodei on stage at the Code with Claude developer conference on Wednesday, May 6, 2026 in San Francisco. (Don Feria/AP Content Services for Anthropic)


On June 10, Anthropic founder and CEO Dario Amodei published a 

 titled, "Policy on the AI Exponential.” He wrote that in recent months “the evidence of AI’s incredible power, as well as its risks, has become undeniable.” He pointed to his company’s product, Mythos, as the "emblematic example” of the threat that frontier models pose to national security. “We need to activate a slow and rickety policy apparatus to deal with risks and opportunities that are going to compound surprisingly quickly from here,” Amodei wrote. He called for “mandatory testing by a qualified third party” of model risks in four areas: (1) cybersecurity, (2) biological weapons, (3) loss of control of AI systems, and (3) “automated R&D that could accelerate these other risks.”

Just two days later, on June 12, Anthropic 

 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.

, concerns raised by Amazon CEO Andy Jassy to the White House instigated the Friday letter. The government, Anthropic said, “believes it has become aware of a method of bypassing, or ‘jailbreaking’ Fable 5,” though the company said the risks were no different than with other models already on the market as it disputed the notion “that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.”

 the episode “one of the most powerful examples yet of US government intervention in the AI race.” Former Trump administration AI advisor Dean Ball 

 the move “baffling,” pointing to the apparent inconsistency between the administration’s willingness to loosen export controls on advanced chips to China and the blanket restriction on access to Anthropic’s advanced models for all foreign governments, including close allies, many of whom 

 on Mythos to identify cybersecurity vulnerabilities in critical systems.

Given these events, observers at the intersection of AI policy and security matters may feel a sense of whiplash. That’s particularly so given this latest dispute comes on the heels of recent policy announcements by the Trump administration that many regarded as a change in tune from the aggressively laissez-faire rhetoric officials advanced a year ago.

Regardless, the immaturity of the federal government’s approach to AI is now on full display. It is likely to have significant and reverberating implications for ongoing debates – and efforts to address – issues at the intersection of AI governance, technological “sovereignty,” and the relationship between US state and corporate power that are playing out in Washington and in capitals all over the world. It could be fairly argued that these ad hoc and wild swings on the part of the US administration are no way to exercise state power, and is further evidence of the need for a regulatory system that provides a more stable equilibrium for stakeholders to operate.

On day four of his second term in office, President Donald Trump issued an 

 titled, “Removing Barriers to American Leadership in Artificial Intelligence.” It rescinded Biden era policy, including a 

 on “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence,” which the Trump order characterized as an obstacle to “global leadership in artificial intelligence.” The order effectively reset the federal government’s AI policy process. In doing so, it discarded Biden-era tools that were designed to surface the types of risks the Trump administration is reacting to post-Mythos, including a scheme for the Department of Commerce to collect red-team results for dual use foundation models.

 of actions and orders that, together, revealed a more complicated picture of the administration’s AI policy. As Alondra Nelson argued in 

 earlier this year, it became clear that the administration's approach involved "not the absence of AI regulation but its rearrangement," operating through industrial policy, trade restrictions, immigration controls, and strategic preemption rather than only traditional regulatory processes. And yet it is often difficult to locate a coherent strategy within that rearrangement. The same administration that sought to strip away oversight in January 2025 is now, in June 2026, fumbling toward using ever more drastic levers of state power to exert control over the AI industry, and Anthropic in particular.

 on whether to release it, on June 2, the administration 

 an executive order "Promoting Advanced Artificial Intelligence Innovation and Security." The order mandates actions to integrate AI into national security infrastructure and to set up a framework for “voluntary” federal oversight. To 

, the order was a step in the right direction, even if—as Mariana Olaizola Rosenblat argued in 

—it “drew the right map” but “stopped at the trailhead” by declining to make the framework binding. Council on Foreign Relations analysts argued the executive order 

 “an administration trying to sustain its deregulatory, innovation-first posture while confronting the novel cyber risks posed by powerful new tools like Anthropic’s Claude Mythos Preview.”

Three days later, the administration released what it 

 a “historic directive” on AI and national security. The 

National Security Presidential Memorandum

 for the government to issue new policies, develop new strategies, and update complicated documents such as the 

 on autonomy in weapons systems. NSPM-11 contains a provision asserting that “the national security enterprise shall ensure, through contractual clauses or other means, that no commercial entity or adversary possesses the capability to prevent use of, disable or degrade, or materially modify without Federal Government knowledge and approval, an AI system that our men and women depend on for their missions.”

Amos Toh, a senior counsel in the Brennan Center's Liberty and National Security Program, told me the NSPM-11 presents as a “study in conflicting impulses,” including the “unresolved tension between the desire to onboard as much commercial, proprietary AI technology as possible despite fear over any constraints or other choices contractors might bake into their products.” But now the government has compelled Anthropic to do to allied governments precisely what the memo insists no vendor should be able to do to the United States: cut off access to a model that is important, if not vital, to their security.

Some point out that Anthropic’s Amodei has played a substantial role in raising the temperature in Washington over AI threats. Former Meta AI head of research Yann Lecun 

 on Facebook that Amodei’s “ridiculous fear mongering” is to blame for these current events. “One reaps what one sows,” he wrote. Adam Thierer, a policy analyst at the R Street Institute, 

 that Anthropic has “relentlessly raised the regulatory temperature in Washington by inviting far-reaching controls of frontier models.” But, he wrote, the administration’s action on Friday nevertheless represents “a significant escalation in the politicization of AI and centralization of control over advanced computation in this country.”

This set of circumstances is likely to raise alarms in foreign capitals about the reliability of American AI for critical applications. Putting no formal, binding policy framework with some kind of legible process in place while being willing to shut down a model 

 suggests the United States will operate ad hoc, potentially with access to advanced models revoked on a case by case basis. To make matters worse, under this administration in particular, the climate is one of a cloud of suspicion that senior officials are picking favorites based on personal and political factors.

Tensions over “tech sovereignty” were already at a boiling point, with policymakers from Canada to Europe taking actions to reduce reliance on American technology and build alternatives. The issue is likely to be an undercurrent this week at the G7 Summit in France, as my colleague Mark Scott points out in 

. Allies “are fearful their countries or regions will become increasingly dependent on already-dominant US tech giants that have shown a willingness to bend to the White House’s political objectives over those of long-time allies,” he writes.

 precisely these sorts of fears over the move to restrict access to Fable and Mythos. Bruno Retailleau, a former French interior minister and now a presidential candidate, wrote 

 that “a nation that depends on others for its technology is a nation that can be unplugged overnight.”

In his June 10 blog post, Amodei called for activating “a slow and rickety policy apparatus to deal with risks and opportunities that are going to compound surprisingly quickly from here.” This month’s whirlwind events suggest “rickety” may be generous. The White House’s improvisation on what appears to be the first real test of what to do about a frontier model’s risks may ultimately imperil its stated goal to ensure American leadership in AI.

Audio of this conversation is available via your favorite podcast service.

In the United States, polling shows a broad bipartisan appetite for AI regulation. An Annenberg Public Policy Center 

 conducted earlier this year found 65% of Americans say the government has done too little to regulate AI, including 77% of Democrats and 53% of Republicans.

On June 4, Reps. Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.) 

 of a bill called the Great American Artificial Intelligence Act of 2026. On the same date, they 

 in Bloomberg Law, explicitly calling for feedback on the draft. “This discussion draft isn’t a final product,” they write. “It’s the start of a serious national conversation with workers, researchers, startups, frontier labs, educators, civil society, state leaders, and the American people.”

, including ”Frontier AI Governance,” “Workforce,” “Cybersecurity,” and “Research, Development, and International Cooperation.” It would be the first comprehensive federal AI governance regime in the United States. If it became law, the bill would create binding federal development obligations for “large frontier developers,” defined as companies with $500M+ in annual revenue that have trained a frontier model. This threshold would capture firms including OpenAI, Anthropic, Google, Meta, and xAI. As analysts at the law firm DLA Piper point out, these provisions appear to draw on frontier model laws developed in 

The draft addresses a range of other issues, including transparency and whistleblower protections. It takes on AI fraud deterrence and cybersecurity. It has a section on “free speech” that would mandate a study and legislative recommendations on jawboning of AI firms and censorship. It supports workforce development and research to better understand the impacts of AI on the labor market. The fourth title of the bill contains provisions related to research and development, “public data for artificial intelligence systems,” and international cooperation on AI governance, which would include advancing American leadership on standards.

 "specifically regulating the development" of AI models on a three-year sunset, while explicitly preserving state laws of general applicability and state authority over activities occurring at or after a model's deployment.

Reaction to the preemption measure drew immediate scrutiny from a variety of civil society organizations and advocates, including 

 Reps. Suhas Subramanyam (D-Va.), Scott Franklin (R-Fla.), Scott Peters (D-Calif.), and Erin Houchin (R-Ind.). Notably, the co-chairs of the House Commission on AI and the Innovation Economy—including Rep. Ted Lieu (D-Calif.), who served as Rep. Obernolte's co-chair on the bipartisan 

House Task Force on Artificial Intelligence

 "cannot serve as the basis for productive dialogue."

To learn more about the draft and their goals for it, I invited Representatives Trahan and Obernolte to the podcast. Rep. Trahan was able to speak to me on Wednesday. She’s been a leading congressional voice on platform transparency, 

 like the Social Media DATA Act and the Digital Services Oversight and Safety Act (DSOSA), which was a similarly ambitious piece of legislation. 

What follows is a lightly edited transcript of the discussion.

My name is Lori Trahan. I represent Massachusetts’ Third District.

Congresswoman, I'm pleased that you're able to join me again on the Tech Policy Press podcast. We've 

 about other legislative proposals that you have put forward, your office and you have become known for taking big swings at big comprehensive solutions to big problems. I think about DSOSA. If we can cast our minds back to the moment when you put that out, a very ambitious, comprehensive plan, particularly around social media. Now you're swinging at artificial intelligence literally with something called the Great American Artificial Intelligence Act of 2026 with your colleague across the aisle, Representative Jay Obernolte. Can I just ask for a second to step back maybe from the specifics of this and ask if in your mind you have a concept or vision of what artificial intelligence means for the United States, what we need to do in order to properly regulate this technology?

Yeah. Well, I think big picture. Artificial intelligence is going to be one of those transformative technologies that changes everything about our life and how we live, how we work. And there's enormous promise that comes with that. I come from Massachusetts, so there's AI being deployed or thought of in the cure and the treatment space for healthcare. But I think what Mythos just demonstrated in an instant is it comes with enormous risk. What I get frustrated by as a member of Congress is when we don't keep up with things that we know are progressing on an accelerated path, it's like Congress doesn't change its ability to pass laws or to establish any governance. And we've seen this before. I've always taken the foundational step of putting transparency reporting in this case auditing because I feel like you can embark on a whole host of other policy areas once you know or once you have visibility into these models, these algorithms, data, whatever it might be, which is why we worked with DSOSA early on in our career and now working on artificial intelligence.

So look, I think AI is going to be the most consequential technology of our generation, and I don't think that's an exaggeration. There's still time to govern it democratically before we seed the future entirely to the private sector or to our foreign adversaries. And that is the bet that we're making with this bill.

Of course, this comes out as a bipartisan proposal. You've obviously had to make multiple trade-offs going into this. The most key one I would reckon is around preemption. A lot of folks in your own party unhappy with this particular approach. Is there a case for preemption that you would make here? What won you over on being willing to sign on to this preemption mechanic in this particular proposal?

So I'm going to answer that question, but I just want to start just like I started with my negotiation with Congressman Obernolte. We didn't start with preemption. And so I'm just going to level set for your listeners and really start by saying something directly to them. We published this bill as a discussion draft because we genuinely want to hear from folks like you. The researchers, the advocates, the civil society groups in this audience, that's exactly who this process is designed for. And this is an opening bid, not a final answer. This draft is comprehensive. It's preliminary and it's narrowly tailored all at once. It's not the one AI framework to rule them all. What it is a focused effort on is tackling catastrophic risks and near term workforce impacts from frontier AI development without strangling the innovative potential of this technology.

And it is designed to equip the federal government with the reporting infrastructure and the technical capacity it needs to govern AI with evidence-driven policy in the years ahead. And so I know we'll get into talking about the specifics of the bill and the governance structure that we put in place, but I'll go to the issue of preemption. I've heard the criticism on the preemption language in the bill and I take it seriously. Look, I've spent years fighting for the exact kinds of state level accountability mechanisms that people in your audience are worried about losing and I'm not going to dismiss that concern. But here's my bottom line on preemption, and I'll just say it plainly. I will preempt, but only if we're setting the strongest possible federal standard.

A preemption provision in a bill with no meaningful federal standard is a moratorium, and I won't sign onto that. So what is the federal lane? Generally speaking, frontier model development, not products, not deployment, not use. The model itself, the weights and architecture trained on petabytes of data on hundreds of thousands of chips, that's the layer where a single federal standard makes sense. The red line that I've held throughout the entire process, states retain full authority to regulate AI at the deployment and the use stage. That is where AI systems actually interact with real people, whether it's housing decisions, hiring, healthcare, credit, states keep their teeth where enforcement is most tangible. The guardrails and the preemption provision are non-negotiable. The savings clauses explicitly preserve laws of general applicability, privacy, consumer protection, anti-discrimination, civil rights, the common law in full. So the cases against Meta and OpenAI proceed and any law regulating activities that occur upon or after a model is released, including the design of AI systems, their deployment and their use by end consumers.

And critically, there's a sunset. Preemption expires in December 2029 assuming passage later this year and that's designed to be a forcing function that prevents Congress from doing what we did with Section 230. And that's implementing a policy and then never being required to seriously evaluate whether it's working. And the sunset forces us back to the table. Now, of course, I also expect Congress to continue to regulate Frontier AI in the years to come, even while the preemption is in effect. These are initial policies that represent a starting point, a foundation for future regulation, I think they preserve optionality as the technology quickly evolves.

Rep. Jay Obernolte (R-Calif.) arrives for a vote at the US Capitol on June 8, 2026. (Francis Chung/POLITICO via AP Images)

Listening to some of those folks that you referenced who've already come out suggesting their concerns around preemption in particular. I think they're really concerned about where the line is drawn when a state wants to regulate a AI companion app that's harming minors. Does it have to regulate the deployer, leave the model developer untouched? Where is this line from your perspective? How do you believe that state law would function under this bill?

Yes. And we want it to. Look, this is a difficult balance to strike and I'll be the first one to acknowledge that. If there are genuine ambiguities in the text that could swallow state consumer protection law or civil rights enforcement, that's a problem I want to fix. That's literally why we published a draft. Come tell us where the language fails. Drawing a clean line between AI model development and AI systems built on those models is genuinely difficult to do in legislative text. I know that, but I don't accept the argument that it's impossible. Laws across the country already leverage the developer deployer distinction. The EU AI Act has specific regulations for general purpose AI models. Legislators draw hard lines every day. That's the job. My sunset combined with the savings clauses is specifically designed to correct for the places where we draw that line imperfectly.

And I'll just bring up one more thing. If you tell me the transparency auditing and verification provisions don't go far enough, just name me the additional policies. We will take them under serious consideration. What I won't accept are mechanisms that are either prone to First Amendment challenges or that are simply bad policy dressed up as accountability. But this is a moment where we want to hear from folks so that we can make the product better.

What do you regard as this period, this feedback solicitation you're doing? You've got this out as a discussion draft, you want to have this dialogue. What do you see as the revision path and are there specific activities or engagements that you're planning that would help along that way?

So first and foremost, it's only been out for a week now. We already have collected a lot of great feedback and with our release we put the email address where people can contact us so that we can start to distill that. It obviously falls in a couple of key buckets, but we encourage more creative ideas and thinking around this. And then we'd love to do round tables. Probably the week after next when we come back into session, we'd love to surface some of the feedback that we've received and have an open dialogue on the ones that we are trying to incorporate into legislative text and maybe the ones where we feel as though the text already covers or whatever it might be, but we want to have an open and transparent process and certainly work with our Republican co-lead on getting the bill into a better position with a coalition before we file it.

I don't put definitive deadlines around that because we really want to make sure that we're keeping it open for as long as we're getting that feedback.

I want to ask just about the broader political context in which this is happening. We're seeing poll after poll that suggests extraordinary concern among the American public really cross political aisle about artificial intelligence, more and more concerned about the infrastructure required to run artificial intelligence, the data center battles that we're seeing spread out in communities across the country, including in Massachusetts. There's a set of fears that the public have about the intent of the corporations and in fact, I think also the intent of the federal government around this. What do you make of this political moment how a bill like this would either will calm or potentially add to that discourse?

Look, I think that people are fearful of how AI is going to disrupt or maybe displace them in a career that they've spent years in. And there's also just a fear of the unknown. But when I'm back home, folks that I talk to, they want us to take a serious effort on this. They know instinctively it needs to be bipartisan. They know that it can't fall victim to the gridlock that has defined Washington for too many years. They're appreciative that I'm working on this. They know that I'm not a shill for big tech. They know that the risk of doing nothing or Congress falling into the traps of gridlock or too much divisiveness will lead to the same thing that they saw with social media. And that's these big, large, mostly private companies setting the rules of the road. And that is, I think, what they would find unacceptable.

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe