Trump just found the worst way to regulate AI
TL;DR
Trump effectively forced Anthropic to take Fable offline after the administration treated a reported vulnerability as an export-control issue. Fable is the heavily restricted public version of Mythos. Anthropic says it used thousands of testing hours, outside reviewers and government checks before release. The move looks selective: according to Vox, experts see the reported jailbreak risk as something many frontier models share, not a unique Fable flaw.
Nauti's Take
This is not the strong AI oversight safety advocates keep asking for. It is the worst version: a powerful intervention without a clean rulebook.
Fable may be risky, no question. But if the government does not have to explain why this model is blocked while similar systems stay online, it is not building a safety regime.
It is building leverage over inconvenient providers.
Briefingshow
AI regulation needs strict safety rules, but also predictable procedure. If a president can block models without transparent standards, political leverage replaces expert review. For developers, companies and foreign customers, that makes US AI less reliable.
Tweets
Sources