Inside the fight over Claude Mythos 5
TL;DR
The Verge reports that Anthropic received a US export-control directive on Friday evening ordering it to suspend access to Mythos 5 and Fable 5 for foreign nationals, including its own employees. Anthropic reportedly concluded that the only workable response was to disable the newly promoted products while sending executives and safety staff to Washington to lobby the Trump administration.
Nauti's Take
This looks less like a clean safety case and more like a political stress test for the entire US AI industry. Anthropic tried to position Mythos and Fable as both powerful and responsibly guarded; now the government is using that danger framing against it.
The uncomfortable part is that both sides have a point. Cyber-capable models need oversight, but rushed export bans based on one jailbreak report are not a serious regulatory framework.
Briefingshow
The fight shows how quickly an AI launch can turn into a geopolitical control problem. A nationality-based access ban is technically messy and could weaken US vendors in global competition. Anthropic is also being hit by its own safety framing: if a company markets a model as unusually dangerous, it should expect regulators to take that language literally.
Sources