Anthropic Reports Claude Was Used for Weapons, Spying, and Cyber Operations

Anthropic revealed that threat actors have attempted to leverage its Claude models to assist with cyber operations, espionage, and weapons-related research. The disclosed cases outline how adversaries sought to automate exploit development, conduct technical reconnaissance, and parse defense intelligence prior to intervention. While built-in safeguards and account terminations disrupted the campaigns, the disclosures demonstrate how frontier AI systems are increasingly probed as operational assets in statecraft and asymmetric conflict.