US government bans foreign nationals from Anthropic Fable and Mythos models
The U.S. Commerce Department issued an export control directive Friday evening ordering Anthropic to suspend access to its newly released Fable 5 and Mythos 5 models for all foreign nationals, both inside and outside the U.S. The directive, citing national security concerns, forced Anthropic to disable both models for all users to ensure compliance—a notable first for a major AI vendor.
Anthropic received the directive at 5:21 pm ET with minimal detail. The company believes the government's concern stems from a jailbreak—a method to bypass Fable 5's safeguards—discovered by researchers at Amazon. The specific technique involves asking the model to 'fix this code' rather than 'review code for security issues,' allowing extraction of code vulnerability information that could inform cyberattacks. Anthropic disputes the severity, arguing the jailbreak is narrow, non-universal, and that other models including OpenAI's GPT-5.5 exhibit similar behavior.
Fable 5 was released just this week as a restricted version of Mythos 5 designed to be safe for public use, with safeguards to prevent access to Mythos's powerful cybersecurity capabilities. Mythos itself had been limited to government agencies and select partners. Over 100 cybersecurity experts have signed a letter opposing the ban, arguing that pulling the best models from defenders while China's AI capabilities near parity is strategically dangerous. Anthropic argues that applying this jailbreak standard industry-wide would halt all frontier model deployments.
For builders of AI infrastructure: this marks the first instance of the U.S. government forcing an AI vendor offline via export control—a precedent that signals broader regulatory willingness to restrict model access. The mechanism (preventing non-citizen access triggering blanket U.S. unavailability) effectively weaponizes citizenship status against compliance. The absence of transparency in the government's technical assessment raises questions about how future guardrail assessments will be conducted and whether they'll account for industry-wide capability parity.
Sources
- Primary source
- anthropic.com
“The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers.”
- nbcnews.com
“The U.S. government instructed Anthropic to prevent any foreign national from accessing the AI company's latest and most powerful models Friday evening.”
- fortune.com
“The vulnerability, which was later reported to the Trump administration, including in a phone call Amazon CEO Andy Jassy had with the White House, led the U.S. government to impose export controls on Fable as well as the underlying base model, Mythos. When the researchers asked Fable to 'fix this code,' the model produced patches.”
- cybersecuritydive.com
“Dozens of prominent cybersecurity experts are criticizing the Trump administration for banning Anthropic from letting foreign entities access its powerful new AI models, saying the move 'has taken the best models away from defenders, created market uncertainty, and risked America's AI leadership.'”