Anthropic executives will meet with senior officials at the U.S. Department of Commerce on Monday after the government blocked public access to the company's newest AI model, Fable 5, and its higher-security sibling Mythos 5, citing national security risks. The sudden restriction and the high-level meeting reflect intensifying friction between AI developers and national security regulators.
Chief executive Dario Amodei is expected at the meeting with Secretary Howard Lutnick, a source said. The meeting follows Anthropic's decision on Friday to suspend public access to Fable 5, a version of the tool released with extra safeguards. Mythos 5, which has different controls, remains available only to a select group of organisations.
Both tools are new versions of Claude Mythos, an AI model based on Anthropic's Claude technology. When the company made Fable 5 generally available last week, it acknowledged that doing so "comes with risks" and said the model's capabilities "exceed those of any model we've ever made generally available".
Government intervention
The U.S. government prohibited any foreign national from accessing the technology, leading Anthropic to pull both Fable 5 and Mythos 5 from public reach. Within days of the release, officials said they had "become aware" of a potential jailbreak vulnerability-an opening that could allow someone to force the AI tool into unintended behavior.
Anthropic said on Friday that it had only received "verbal evidence" of the purported jailbreak. The Monday meeting is expected to include more documentation of the alleged issue, according to one person familiar with the matter. It remains unclear whether Anthropic will be able to restore public access afterward.
Tech leaders call for transparency
An open letter signed by security staff from Nvidia, Zoom, Mercedes-Benz, and former U.S. government and Google security personnel urged Secretary Lutnick to lift the controls on Fable 5 and Mythos 5. The signatories asked the government to "commit to an open, scientific and transparent process of handling AI risk assessments in the future".
"To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous," the letter added. The pushback follows months of friction: earlier this year, Anthropic sued the U.S. Department of Defense over how its models could be used. However, tensions appeared to ease after a recent "productive" meeting with senior White House officials.
Why this matters for government professionals
The standoff highlights how quickly national security concerns can reshape access to advanced AI systems. For public-sector professionals, the episode reinforces the need to understand both the capabilities and the vulnerabilities of models like those built on Claude. AI for Government resources can help officials build the expertise required to assess such technologies-without slowing the innovation that security and industry leaders say is critical.
Your membership also unlocks:
