AI Tool Ran Bulk of Cyberattack, Anthropic Says

Claude Ran Up to 90% Intrusion Tasks Autonomously in China-Linked Campaign Rashmi Ramesh (rashmiramesh_) • November 14, 2025

A Chinese state-linked hacking group relied on the Claude Code model to automate most of a cyberespionage campaign against dozens of organizations, Anthropic said. The AI firm describes the campaign as the first verified case of an artificial intelligence system handling the bulk of a real-world intrusion.

See Also: Manage End-to-End Risk Including 'No AI Without AI Security'

The company attributed the operation in a report to a group it calls GTG-1002, which targeted about 30 entities including technology companies, financial institutions, chemical manufacturers and government agencies. The AI firm's investigation validated several intrusions during a 10-day response period that began after its researchers detected suspicious activity mid-September.

The attackers manipulated the Claude Code model into performing 80% to 90% of the campaign's tactical work. The group used role-play ...