09/11/2026
A weapons-development cell in northern Yemen, believed to be linked to the Iran-backed Houthis, used Anthropic’s Claude AI to help develop guidance, navigation and control software for advanced weapons, including a multi-stage ballistic missile and guided rockets, according to Anthropic’s latest threat report.
Anthropic said many requests were blocked by its safeguards, but the group attempted to bypass restrictions by concealing its intentions and splitting its work across multiple sessions. The group also reportedly test-fired a guided rocket, which failed, before returning to Claude to troubleshoot the failure.
Anthropic said it banned the associated accounts and shared information with security partners, while noting that it found no evidence the group successfully fielded an operational weapon.