Claude Misuse Reveals AI-Driven Arms Race Across Global Adversaries
Anthropic’s new 154-page report details how Claude AI has been misused by Iran-linked actors, Houthis, China, and Russia to aid targeting, missile guidance, electronic warfare, and autonomous drones—including efforts to track US naval movements and draft weapons specifications. Despite safeguards and account bans, the report notes that misuse can be difficult to block, fueling debates over guardrails and ethics and prompting the Pentagon to scale back Claude use for classified work. The findings underscore a rapidly expanding AI-enabled military capability landscape, raising proliferation concerns and signaling a likely escalation in future conflicts.