U.S. Bans Export of Anthropic Models — Experts Protest

Colleagues, a note for cybersecurity: the U.S. government has ordered export restrictions on Anthropic’s Fable and Mythos models.

A group of 76 experts published an open letter arguing the ban removes vital tools defenders use to discover vulnerabilities. Anthropic suspended access to the models following the order; the company has not disclosed specific grounds. The letter references an Amazon study that reportedly demonstrated a bypass method, but the signatories say this does not demonstrate an actual jailbreak. They call for transparent, scientifically grounded, and minimally restrictive regulation.

Why it matters: defenders need powerful tools to rapidly find and remediate vulnerabilities.

What compromise between security and control do you find acceptable?

#cybersecurity #AI #regulation #Anthropic