Cybersecurity vets protest ‘dangerous’ US government ban on Anthropic’s most powerful models
TL;DR
76 cybersecurity experts are urging the US government in an open letter to remove export restrictions on Anthropic’s Fable and Mythos models. The order arrived Friday on national security grounds. Anthropic responded by suspending access to both models worldwide, not only for users outside the US. The signatories argue the move mainly hurts defenders: Fable and Mythos can help find vulnerabilities, write patches, and generate tests.
Nauti's Take
The open letter campaign has clear self-interest, but its core argument lands: defenders need access to the same strong tools attackers will test anyway. A worldwide access freeze is politically simpler than a narrow, evidence-based rule, but it is a shaky security move.
Anthropic positioned Mythos as unusually powerful; now that risk messaging has turned into a regulatory problem.
Briefingshow
This is a test case for how governments restrict dangerous AI capabilities without weakening defensive security work. A model that can find bugs is not automatically an attack tool. Without a transparent rationale, the measure looks blunt and pushes companies, researchers, and bug bounty teams toward weaker or less controlled alternatives.
Sources