Feds’ legal basis for ban on Anthropic’s most powerful models looks increasingly shaky

At a glance:

• The Trump administration banned Anthropic’s Fable 5 and Mythos 5 models five days ago, citing national security concerns.
• Legal experts argue the ban misuses export regulations and lacks clarity on restricting API access.
• The move has strained U.S. relations with allies like France and the UK, raising questions about AI governance.

The sudden ban and its immediate fallout

The Trump administration’s abrupt ban on Anthropic’s most advanced AI models—Fable 5 and Mythos 5—has thrown the AI industry into uncertainty. On Friday, Secretary of Commerce Howard Lutnick sent a letter to Anthropic CEO Dario Amodei demanding the company halt use of these models by "foreign persons" within 90 minutes, a deadline that left Anthropic scrambling to cut off all user access. The directive, framed under the Department of Commerce’s Export Administration Regulations (EAR), claims the models pose risks of "military-intelligence end use" or "end user." However, the letter omitted any mention of cybersecurity concerns, despite reports that Amazon CEO Andy Jassy had briefed Treasury Secretary Scott Bessent on Anthropic’s research revealing vulnerabilities in Amazon’s own cybersecurity software. Amazon, a major investor in Anthropic and provider of GPUs for its Claude models, has yet to publicly address the ban’s implications for its partnership.

Legal experts question the administration’s authority

Cybersecurity and legal experts have swiftly challenged the ban’s rationale. An open letter signed by dozens of professionals argued that Fable 5, which was publicly released, includes robust safeguards and that identifying cybersecurity flaws is not unique to Anthropic’s models. Many signatories noted they regularly use other foundation or open-source models for security audits. Alasdair Phillips-Robins, a former Department of Commerce adviser, criticized Lutnick’s letter as poorly drafted, stating that API access does not constitute an "export" under EAR. He warned that the administration’s approach could set a precedent for broad restrictions on AI development, effectively criminalizing the release of advanced models by U.S. companies. Phillips-Robins emphasized that if the White House views certain AI models as threats, it should pursue legislative action rather than relying on existing export laws.

International fallout and G7 tensions

The ban’s global scope—applying even to U.S. allies like France and the UK—has sparked diplomatic friction. During the G7 Summit in Évian-les-Bains, French President Emmanuel Macron met with Amodei and OpenAI CEO Sam Altman to discuss European access to cutting-edge AI models. The timing raised eyebrows, as the ban was announced just days before the summit. President Trump reportedly denied UK Prime Minister Keir Starmer’s request for an exemption, signaling a hardline stance. European leaders are increasingly anxious about overreliance on U.S. AI technologies, with the ban exacerbating fears of being cut off from critical innovations. The move could reshape global AI partnerships, forcing allies to seek alternatives or negotiate new terms.

Regulatory uncertainty and industry implications

The ban highlights a broader shift in the Trump administration’s AI policy. Historically, officials favored a hands-off approach, but recent reports suggest growing interest in leveraging export laws to control AI distribution. This pivot raises concerns about regulatory overreach and its potential to stifle innovation. For Anthropic, the ban underscores the risks of operating in a politically charged environment where national security claims can override commercial and technical considerations. The company’s blog post defending Fable 5’s safeguards and warning against blanket restrictions reflects a growing tension between AI developers and governments grappling with oversight.

What comes next for Anthropic and AI governance

Anthropic has not yet indicated whether it will challenge the ban in court, leaving the legal battle’s trajectory unclear. Meanwhile, the administration’s reliance on EAR sets a precedent that could extend to other AI firms. Legal experts stress that without congressional action, such measures may lack durability. The incident also underscores the need for clearer frameworks governing AI safety and international access. As nations like France push for greater autonomy in AI development, the ban may accelerate efforts to diversify supply chains and reduce dependence on U.S. tech giants.

Conclusion: A test case for AI regulation

The ban on Fable 5 and Mythos 5 represents a pivotal moment in AI governance, blending cybersecurity debates, international diplomacy, and regulatory ambiguity. While the administration frames it as a national security imperative, critics argue it risks undermining trust in U.S. AI leadership. The outcome will likely influence how governments balance innovation with oversight, and whether companies like Anthropic can navigate an evolving landscape of geopolitical and legal pressures.