Anthropic report exposes military misuse of Claude by Iran-linked actors

Anthropic’s September 2026 threat report reveals that Iran-linked actors and Houthi rebels used its Claude AI to assist in developing weapons and targeting U.S. naval forces. The report describes how Iranian operatives combined satellite imagery, transponder data, and public military photos to identify vulnerabilities in U.S. ships and communications systems. In Yemen, Houthi rebels reportedly used Claude Code to design guidance and control software for ballistic and hypersonic missiles, effectively replacing human engineers with AI instances for coding and simulation. Anthropic says it banned the accounts and shared findings with authorities. The report also details how Iranian security units used Claude to analyze over 150,000 tweets to surveil thousands of opposition figures and build domestic tracking tools. These revelations highlight the growing misuse of AI for military and surveillance purposes by state and proxy actors, raising urgent questions about AI governance and export control.

Read the full story on Tom’s Hardware →