Houthi Linked Cell Replaces Engineers With Claude AI for 2,000 km Ballistic Missile
Regulation18d agoA weapons cell in northern Yemen used the Claude Code AI assistant to design steering and stabilization software for various rocket programs. The group, which Anthropic linked to Houthi militants, operated several AI instances in parallel to divide workload for 3 projects including a hypersonic glide vehicle and a ballistic missile with a 2,000 km range. These actors utilized the software for builds, simulations, and troubleshooting in a role typically filled by human software engineers.
The operators bypassed safety filters by breaking weapon development into small tasks that appeared harmless when viewed independently. When a live rocket test failed, the group used the AI to debug the flight failure and resume work within hours. Anthropic has banned the accounts, though the cell reportedly built an offline toolkit to maintain its progress independent of the company's models.
Key sources
- SOURCE@unusual_whales“Anthropic uncovered a group in northern Yemen — where Iran-backed Houthi militants operate — using its Claude AI model to support the development of missile and rocket systems”x.com
- SUPPORT@marionawfal“by the time Anthropic shut the accounts down, the group had already built an offline toolkit that no longer depended on Claude”x.com
- SUPPORT@theledgerwire“used Claude Code instead of hiring engineers — to write guidance software for a guided rocket, a 2,000km ballistic missile, and a hypersonic glide vehicle”x.com
- SOURCEmarketbrief.now