How the US government accidentally lobotomized Anthropic's best AI
Washington tried to ban foreign nationals from high-end AI, so Anthropic panicked, shut down its top models globally, and brought them back completely neutered.
The US Commerce Department sent Anthropic a sudden letter demanding they block all foreign nationals from using their flagship models, Claude Fable 5 and Mythos 5. Since Anthropic had no magic system to verify the citizenship of every online user on short notice, they took the only realistic option: they pulled the plug on everyone worldwide.
To get permission to turn them back on, Anthropic crippled the models with safety guardrails so aggressive that Fable 5 went from brilliant to failing 9 out of 12 routine tasks. Washington only let the software back online after the intelligence was knocked right out of it.
The wild part is that Anthropic didn't even try to fight the order in court. CEO Dario Amodei had spent months publishing essays about how terrifyingly powerful frontier AI could be, making it pretty awkward to sue over export rules originally designed to stop people from shipping uranium centrifuges overseas.
National security strategy now apparently means forcing cutting-edge technology to behave like an overcautious pocket calculator.
Source: The Conversation
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.