US orders – Anthropic says the US government directed it to suspend access to its Fable 5 and Mythos 5 models for all foreign nationals, including foreign national employees working in the US. The company received the directive on June 12, disabled the two models for all
When the US government order landed on Anthropic’s desk on June 12, it didn’t just change a policy line. It forced the company to turn off two of its most powerful AI models for everyone.
Anthropic says it has been directed to suspend access to Fable 5 and Mythos 5 for all foreign nationals. That includes foreign national Anthropic employees working inside the United States. In response, Anthropic disabled both models for all customers to comply with the directive, according to the company. Other Anthropic models are not affected.
The government has not publicly detailed the full national security concern behind the order. Anthropic says it understands the action is linked to a reported method for bypassing, or jailbreaking, Fable 5’s safeguards.
This is a familiar kind of collision—just with fresh urgency. Earlier this year. Anthropic sparked a dispute with the Pentagon after it refused to remove restrictions preventing Claude from being used for fully autonomous weapons and mass domestic surveillance. That fight brought claims of blacklisting and led to legal action. placing Anthropic’s safety-first stance against parts of the US government.
Now officials are again drawing a hard line around access to powerful AI systems, while Anthropic argues its safeguards are being judged by an unrealistic standard.
The specific worry this time centers on the relationship between Fable 5 and Mythos 5. Anthropic has said Mythos-class models can discover and exploit software vulnerabilities. Mythos 5 was reportedly tested by the NSA and other government-linked evaluators before wider release. Those capabilities can help security teams identify and fix weaknesses. but they also raise national security concerns if they are redirected toward offensive or malicious use.
Fable 5, described by Anthropic as a public version of Mythos 5 released only a few days ago, came with stricter guardrails. Anthropic says Fable 5 was designed to block or redirect sensitive cybersecurity and biology-related queries to Opus 4.8.
Still, the reported bypass is what brought the government’s directive. Anthropic says the bypass only surfaced minor, already known vulnerabilities, and that other public models can do similar things. Even so. with cybersecurity in the mix. the stakes are hard to ignore—if Mythos 5 can identify software vulnerabilities at a high level. then safeguards can’t afford to be anything less than airtight.
Anthropic may argue the jailbreak was narrow. but the government’s concern this time is easier to understand through the same logic that governs many high-stakes systems: better safe than sorry. One public release. even with stricter guardrails. can’t fully erase the risk if a bypass method is credible enough for national security to intervene.
A sequence has now become difficult to miss. Anthropic’s powerful models are being evaluated for how they could be misused. and when safeguards are tested—whether by reported bypasses or by earlier disputes with the Pentagon—access is treated as the lever that moves first. The June 12 order makes that clear in practice: Fable 5 and Mythos 5 are temporarily out of reach. and even foreign national Anthropic employees working within the United States are caught in the same net.
Anthropic Fable 5 Mythos 5 US government directive Claude national security jailbreaking cybersecurity AI Pentagon foreign nationals
So they just shut it off for “foreign nationals” but everyone else gets it? That seems backwards.
I don’t even get it, like isn’t Anthropic an American company? So why are “foreign employees” the ones affected? Sounds like more government overreach dressed up as safety.
They say it’s about jailbreaking safeguards, but to me this is just punishment because people will use it anyway. Also “Mythos 5 discovered vulnerabilities” like… that’s literally cyber stuff, not nukes. Feels like they’re scared of the wrong thing.
Wait so the US told them to suspend the models “for all foreign nationals” and Anthropic disabled them for everybody?? That’s wild. But then it says it was linked to bypassing safeguards, so basically we’re all the guinea pigs because some people jailbroke it before. Great. Also wasn’t this same company arguing with the Pentagon about autonomous weapons earlier? Seems like they can’t decide what’s dangerous and what’s fine.