The US government has ordered AI company Anthropic to shut off access to two of its latest AI models, Fable 5 and Mythos 5, citing national security concerns.
The directive, issued Friday evening, forces Anthropic to disable both models for all customers globally, including foreign national employees working inside the company. It’s a sweeping move that caught the AI industry off guard.
Anthropic said the government’s letter didn’t spell out exactly what the national security threat was, but the company understands the order stems from a newly discovered jailbreak, a method to bypass the models’ safety guardrails. After reviewing a demo of the technique, Anthropic pushed back on the severity of the finding, saying it only exposed a small number of minor, previously known vulnerabilities that other publicly available models could already surface without a special bypass.
The company made clear it disagrees with the government’s response. “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. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
Anthropic also pointed out that the capabilities exposed by the exploit are already accessible through competing models, including OpenAI’s GPT-5.5, which makes the targeted shutdown feel even more heavy-handed from the company’s perspective.
The company says it’s complying with the order while working with officials to get access restored. “We apologize for this disruption to our customers,” Anthropic said. “We believe this is a misunderstanding and are working to restore access as soon as possible.”
The sudden loss of access to Anthropic’s Mythos model leaves Canadian cybersecurity initiatives in limbo after Ottawa had just secured specialized entry through Project Glasswing. Mythos was touted as being able to bypass cybersecurity of numerous systems as it exploited old or bad code.
All other Anthropic models remain available and unaffected.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
