Anthropic PBC is seeking to resolve a fresh confrontation with the Trump administration over artificial intelligence security that prompted the company to disable global access to its two most advanced AI models. Senior technical staff from Anthropic met with administration officials at the Commerce Department on Monday to discuss the national security concerns raised by the government, according to a person familiar with the planning. It was not immediately clear which administration officials participated in the talks.
An Anthropic spokesperson said following the meeting that both sides were working quickly to reach a solution. Spokespeople from Commerce did not immediately respond to requests for comment. The discussions come days after the Trump administration told Anthropic to halt foreign nationals’ access to the company’s cutting-edge Fable 5 and Mythos 5 models. Since then, representatives from Anthropic have held numerous virtual meetings with U.S. officials about specific security issues, the company spokesperson said.
Government Intervention in AI Operations
The order, sent from the Commerce Department to Anthropic late Friday afternoon, represents the most significant intervention by the U.S. government to date into an AI venture’s operations. It poses a new challenge to Anthropic weeks after the company filed confidentially for an initial public offering, with its latest valuation topping US$900 billion. In a website post late Friday, Anthropic said it had complied with the U.S. directive by disabling access to the two models.
The company said it believes the U.S. government issued the order after discovering that it is possible to “jailbreak,” or bypass the guardrails, of Fable 5, a recently released version of Mythos that Anthropic had blocked from carrying out cybersecurity tasks. “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” Anthropic said in its post. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
Anthropic's Position on AI Safety
Privately held Anthropic, which has long positioned itself as a more responsible AI developer, first shared its Mythos model in April with a very limited group of companies and institutions, warning that its ability to find cybersecurity vulnerabilities made it too risky to distribute more widely. Last week, the company released Fable 5 as the first public-facing version of its Mythos-class model, but with guardrails aimed at containing the full range of its cyber capabilities.
Ayham Boucher, executive director of AI Strategy and Innovation at Cornell University, says the debate is being framed as a simple request for Anthropic to “fix the guardrails” and then remove the Fable 5 ban, but that underestimates the difficulty in resolving a highly technical problem. The outcome of these talks could set a precedent for how the U.S. government regulates advanced AI models in the future.
