Anthropic has disclosed that actors linked to Yemen’s Houthi movement used its Claude AI coding tool to develop guidance software for rockets and ballistic missiles. One guided rocket was test-fired, though the trial appears to have failed.
In its September 2026 misuse report, Anthropic detailed how a Yemen-based weapons engineering cell used Claude Code to work on three programmes: a guided rocket using a phone-class flight computer with final-phase homing guidance, a multi-stage ballistic missile with a stated range goal exceeding 2,000 kilometres, and a multi-variant missile set that included a hypersonic glide vehicle variant.
The actors used Claude in place of human software engineers to develop guidance, navigation and control software. They ran multiple Claude instances simultaneously, assigning each a specific role — one writing code, another conducting research, a third reviewing the output — replicating the structure of a small engineering team.
Anthropic said its safety systems blocked many of the group’s requests but not all. The actors evaded safeguards by concealing their objectives and splitting work across multiple sessions so no single session revealed the full scope. Anthropic’s internal investigations flagged the activity. The company banned the associated accounts and shared threat intelligence with public and private sector partners.
The company said it found no evidence the actors fielded an operational device. After the failed rocket test, the actors returned to Claude within hours to troubleshoot. Anthropic also noted that the group had already developed an offline simulation toolkit independent of Claude or other engineering computing tools like MATLAB.
The disclosure is significant because it represents one of the clearest documented cases of an AI coding tool being used in an active weapons development effort.
Source: Business Standard
