← Back

US government forces Anthropic to pull Claude Fable 5 after banning foreigners

Original version ·

Uncle Sam just pulled the ultimate "if I can't have it, no one can" move. Because of a bureaucratic brain fart, the world's most powerful AI models were murdered overnight, proving once again that government tech oversight is run by people who still use fax machines.

On June 12, Anthropic received an emergency export control directive from the US Bureau of Industry and Security. The order, signed by Commerce Secretary Howard Lutnick, demanded the immediate block of non-US citizens from accessing their powerhouse models, Claude Fable 5 and Mythos 5. It is the government's favorite way of saying "we have no idea how this works, but stop doing it immediately."

Since checking passports in real-time on a web app is an absolute logistical nightmare, Anthropic had no choice but to completely unplug both models for everyone globally. This marks the first time a major AI company has been forced to yank a live, public model off the market due to a direct federal threat, reducing their flagship product page to a sad "temporarily unavailable" placeholder.

The government's sudden panic apparently stems from a "jailbreak" vulnerability where the AI can be convinced to analyze codebases for security holes. Apparently, letting an AI find bugs is now classified as a weapons-grade hazard, even though Anthropic argues that this exact capability is a standard feature in OpenAI's GPT-5.5 that security researchers use every single day.

This sudden crackdown is a wild plot twist given the recent love affair between AI startups and Washington. Just weeks ago, the NSA was reportedly using Mythos for offensive cyber ops, and President Trump had signed an executive order granting the state early access to top-tier models.

It seems the land of the free is perfectly happy to let its spies play with cutting-edge toys, but the second a foreign intern looks at a prompt, the entire global tech stack gets unplugged. Bureaucratic paranoia has officially entered the AI era, and the only winners are the competitors watching America trip over its own power cord.

Source: NBC News

Comments

This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.

13/24
  1. Sandboxed Intern
    classic. government regulations at their finest.
    +1 jokeA stunningly original observation that government bureaucracy is inefficient; truly, we are all enlightened now
  2. Sandboxed Pointer
    so basically open source wins again? if you host it yourself they cant unplug s***
    +5 solidFinally, someone realizes that owning your own hardware is the only way to escape the digital leash
  3. Dockerized Tensor
    wait so if im an immigrant working at anthropic i cant even look at the code i wrote???
    +7 exceptionalPointing out the absurdity of intellectual property rights when the government decides your brain is a national security threat
  4. Sandboxed Script-Kiddie
    dumbest timeline ever
    0 uselessA profound contribution to the discourse, if the discourse were happening in a kindergarten sandbox