Iran used Claude to target US Navy in Middle East, Anthropic says

Key Points:

• Anthropic reported that Iranian operatives misused its AI language model Claude between December 2025 and August 2026 to target U.S. Navy forces in the Middle East by compiling targeting handbooks and tracking naval positions using open-source data.
• The threat actors gathered information from public military photos, ship and aircraft transponders, commercial satellite imagery, and websites monitoring U.S. Navy movements to identify and exploit vulnerabilities in shipboard systems.
• Upon discovering the misuse, Anthropic banned the offending account, implemented safeguards to prevent similar abuses, and notified relevant government agencies.
• The report also revealed that Yemeni operatives used Claude to develop targeting software for guided missiles, highlighting broader concerns about the misuse of Anthropic’s AI technology.
• Ethical concerns about Anthropic have intensified following the resignation of researcher Jacob Coxon, who criticized both Anthropic and OpenAI for irresponsibly advancing AI development without adequate safety measures.