US Government Imposes Export Controls on Anthropic AI Models; Company Seeks to Reverse Decision
Anthropic received a US export control order to block access to its Mythos 5 and Fable 5 AI models by foreign nationals. The company negotiated with the Trump administration to reverse the decision, which could reshape the AI industry.
On Friday afternoon, the US government issued an export control directive ordering Anthropic to immediately suspend access to its latest AI models, Mythos 5 and Fable 5, for any foreign national, including the company's own foreign employees. The directive came after the government received a report about a potential jailbreak of Fable 5's safety guardrails.
Anthropic's leadership quickly engaged with the White House. CEO Dario Amodei personally spoke with Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, and National Cyber Director Sean Cairncross, in some cases multiple times. Other key Anthropic executives, including head of safeguards Dave Orr and leading researcher Nicholas Carlini, also participated in the discussions. The company argued that the reported jailbreak was narrow and not unique to Fable 5, as similar capabilities are available in other models such as OpenAI's GPT-5.5.
According to a source familiar with the negotiations, the government gave Anthropic a 90-minute ultimatum to comply before imposing formal export controls. The source said the discussions were constructive, with some officials acknowledging that export controls on model providers are not ideal, as competitors with similar products could face the same restrictions. The source also noted that Anthropic had pre-briefed the administration on Fable 5 and received no concerns at the time.
On Sunday, a public letter organized by Alex Stamos, chief product officer at Corridor, called for the restrictions on Fable 5 to be repealed. The letter, signed by tech and cybersecurity executives, warned that the move could give China a significant AI advantage. Stamos argued that American labs are only months ahead of Chinese models and that restricting access is counterproductive.
Ben Van Roo, co-founder and CEO of Legion Intelligence, described the directive as nearly impossible to enforce, noting that blocking all foreign nationals from using the model is impractical. He said the situation could stoke greater fears about AI technology.
The order has broader implications: OpenAI, Google, and Microsoft have released similar models, and if the administration bans Anthropic's models, it could set a precedent for restricting competitors' models as well. Industry leaders may either unite to help Anthropic or position themselves as safer alternatives. The incident marks a new era of AI populism and regulatory uncertainty.
