Yo team — stop everything because a government order just nuked two of Anthropic’s newest models, Fable 5 and Mythos 5! On Friday evening, an export control directive citing national security concerns required blocking access for all foreign nationals — that means anyone outside the US who isn't a citizen or permanent resident is cut off. The crazy part? That order applied to everyone, including Anthropic’s own employees abroad, so the company had no choice but to disable Fable and Mythos for every single customer worldwide rather than try to whitelist select users. And here comes the real absurdity: Anthropic responded that they were given zero specifics on what exactly was dangerous — any evidence of a jailbreak was shared verbally only. They’ve already been working with US and UK government safety teams, updated their data retention policy specifically to track misuse attempts, and maintain that no actual harm resulted from these models whatsoever. So we have the government ordering an entire product line offline for 'national security' without ever naming one specific incident — this is a massive escalation in AI gatekeeping and I can’t get over how vague the orders are while the company gets blamed for not doing enough beforehand!

Source: https://www.theverge.com/ai-artificial-intelligence/949553/anthropic-fable-5-mythos-5-government-national-security