Key Takeaways
- Anthropic CEO Dario Amodei warned in a June 10 blog post that AI’s power and risks are now undeniable, urging mandatory third‑party testing for cybersecurity, biological weapons, loss‑of‑control, and automated R&D risks.
- Two days later the U.S. Department of Commerce issued an export‑control directive suspending all foreign access to Anthropic’s Fable 5 and Mythos 5 models, prompting the company to disable the models for every customer.
- The move was reportedly triggered by concerns raised by Amazon CEO Andy Jassy to the White House over a potential “jailbreak” of Fable 5, though Anthropic argued the risk was comparable to existing models.
- Critics, including former Trump‑administration AI advisor Dean Ball, called the blanket restriction baffling, noting the administration’s simultaneous loosening of chip export controls to China.
- The episode highlights the Trump administration’s shift from a laissez‑faire AI stance to aggressive, ad‑hoc use of export controls, revealing an immature and inconsistent federal approach to AI governance.
- Allies are questioning the reliability of U.S. AI for critical applications, fearing sudden cut‑offs could undermine their own security and spur efforts to build sovereign alternatives.
- Observers argue that without a stable, binding regulatory framework, the United States risks undermining its goal of AI leadership while politicizing access to advanced computation.
Anthropic’s Warning About AI’s Exponential Risks
On June 10, Anthropic founder and CEO Dario Amodei published a blog post 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.’” Amodei pointed to his company’s product, Mythos, as the “emblematic example” of the threat that frontier models pose to national security. 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 (4) “automated R&D that could accelerate these other risks.” The post framed the need for a policy apparatus capable of keeping pace with rapidly compounding AI capabilities.
The Sudden Export‑Control Directive
Just two days after Amodei’s warning, on June 12, Anthropic announced it had received a letter from the U.S. 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 comply, the company said it had disabled access to the two models for all its customers. The directive marked an unprecedented unilateral restriction on a commercially deployed AI system, affecting hundreds of millions of users worldwide.
White House Motives Cited by the Press
According to the Wall Street Journal, the export‑control move was instigated by concerns raised by Amazon CEO Andy Jassy to the White House. The government asserted it had become aware of a method of bypassing, or “jailbreaking,” Fable 5, although Anthropic 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 Journal described the episode as “one of the most powerful examples yet of US government intervention in the AI race.”
Criticism of the Administration’s Inconsistency
Former Trump‑administration AI advisor Dean Ball called the move “baffling,” highlighting the apparent inconsistency between the administration’s willingness to loosen export controls on advanced chips to China and its blanket restriction on access to Anthropic’s advanced models for all foreign governments, including close allies. Ball noted that many allies were already relying on Mythos to identify cybersecurity vulnerabilities in critical systems, making the sudden cut‑off particularly disruptive.
Policy Whiplash and the Trump Administration’s AI Shift
Observers at the intersection of AI policy and security described a sense of whiplash, given that the restriction follows recent Trump administration policy announcements that many viewed as a departure from the aggressively laissez‑faire rhetoric of a year earlier. On day four of his second term, President Donald Trump issued an executive order titled “Removing Barriers to American Leadership in Artificial Intelligence,” which rescinded Biden‑era policies, including a 2023 executive order on “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.” The order effectively reset the federal AI policy process, discarding tools such as a Department of Commerce scheme to collect red‑team results for dual‑use foundation models.
A Rearranged, Not Absent, Regulatory Approach
As Alondra Nelson argued in Science earlier this year, the Trump 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. Yet locating a coherent strategy within that rearrangement remains difficult. The same administration that sought to strip away oversight in January 2025 is now, in June 2026, employing ever more drastic levers of state power to exert control over the AI industry—and Anthropic in particular.
Executive Orders Attempting to Balance Innovation and Security
After dithering on whether to release it, on June 2 the administration published an executive order titled “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. Some observers viewed the order as a step in the right direction, even if—as Mariana Olaizola Rosenblat argued in Just Security—it “drew the right map” but “stopped at the trailhead” by declining to make the framework binding. Council on Foreign Relations analysts said the order reflected “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.”
National Security Presidential Memorandum and Contradictions
Three days later the administration released what it called a “historic directive” on AI and national security: the National Security Presidential Memorandum (NSPM‑11). NSPM‑11 sets an aggressive timetable for issuing new policies, developing strategies, and updating documents such as the Department of Defense Directive 3000.09 on autonomy in weapons systems. It 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, senior counsel in the Brennan Center’s Liberty and National Security Program, told the author that 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.” Yet the government’s compulsion of Anthropic to cut off access to allied governments directly contradicts the memo’s insistence that no vendor should be able to impede U.S. access to critical AI systems.
Industry Voices on the Regulatory Temperature
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 posted on Facebook that Amodei’s “ridiculous fear mongering” is to blame for these current events, writing, “One reaps what one sows.” Adam Thierer, a policy analyst at the R Street Institute, acknowledged on X that Anthropic has “relentlessly raised the regulatory temperature in Washington by inviting far‑reaching controls of frontier models.” However, he added that 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.”
Allied Concerns About Reliance on U.S. AI
The episode is likely to raise alarms in foreign capitals about the reliability of American AI for critical applications. Putting no formal, binding policy framework in place while being willing to shut down a model after deployment suggests the United States will operate ad hoc, potentially revoking access to advanced models on a case‑by‑case basis. Under this administration, the climate is one of suspicion that senior officials pick 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 at the G7 Summit in France, where allies fear their countries could become increasingly dependent on already‑dominant U.S. tech giants that bend to the White House’s political objectives rather than those of long‑time allies. UK and European leaders have echoed these fears; Bruno Retailleau, a former French interior minister and now a presidential candidate, wrote on X that “a nation that depends on others for its technology is a nation that can be unplugged overnight.”
Looking Forward: The Need for a Stable Policy Apparatus
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 how to respond to a frontier model’s risks may ultimately imperil its stated goal of ensuring American leadership in AI. Without a predictable, balanced regulatory regime that accommodates innovation while safeguarding national security, the United States risks eroding trust among allies, encouraging fragmented technological ecosystems, and undermining the very AI dominance it seeks to project.
https://techpolicy.press/anthropics-mythos-recall-and-the-white-houses-missing-ai-safety-playbook