Trump just found the worst way to regulate AI
TL;DR
Vox reports that the Trump administration forced Anthropic to take its new Fable model offline. The move used export controls after a reported jailbreak risk, not a clear AI safety law. Fable is the public, tightly restricted version of Anthropic Mythos. Vox says it refuses sensitive cyber and biology requests, but officials still treated possible bypasses as a security threat.
Nauti's Take
Frontier AI needs strict safety rules, but that makes process even more important. A president using export controls to pull individual models from the market does not create safety; it creates uncertainty for developers, customers, and international partners.
This is the worst mix: enough state power to hit companies, but not enough rule of law to build trust.
Briefingshow
The issue is not that powerful AI models face regulation. The issue is a precedent where the White House can effectively ban a model without clear standards. If rules depend on political access, internal rivalries, or vague security claims, AI safety becomes a power tool instead of reliable oversight.
Tweets
Sources
