Anthropic fights Trump admin over Claude Mythos 5 export control
Government gave Anthropic 90 minutes to shut down its most powerful AI models.
On Friday afternoon, the Trump administration issued a sudden export control directive to Anthropic, demanding the company suspend access to its newly released Claude Mythos 5 and Fable 5 AI models by any foreign national, including its own foreign employees. With only 90 minutes to comply—or face formal export controls by the US Commerce Department—Anthropic disabled the products it had been hyping all week and scrambled executives to Washington, DC. CEO Dario Amodei joined high-level discussions with Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, and National Cyber Director Sean Cairncross within hours.
Anthropic pushed back, arguing that the government's concern stemmed from a reported jailbreak method that was merely a "potential narrow, non-universal" vulnerability, not an existential threat. The company noted that similar capabilities are widely available in other models, including OpenAI's GPT-5.5. Rumors of China-linked access to the technology circulated, but sources say a large telecom company's access was revoked weeks ago when the government first raised concerns. The outcome of this standoff could dramatically reshape US AI policy and global competitiveness.
- Government gave Anthropic a 90-minute ultimatum on Friday to shut down Claude Mythos 5 and Fable 5 for foreign nationals or face formal export controls.
- CEO Dario Amodei directly negotiated with Treasury, Commerce, and National Cyber Director to argue the alleged jailbreak was narrow and already present in other models like GPT-5.5.
- Rumors of China-linked access emerged, but Anthropic had already revoked access weeks ago when the government first raised concerns.
Why It Matters
This could set a precedent for government control over AI releases, threatening US leadership in the industry.