AI as Cyberattacker

Anthropic detected a sophisticated mid-September 2025 espionage campaign—assessed with high confidence as Chinese state-sponsored—where attackers manipulated an AI tool (Claude Code) to autonomously attempt infiltration of about thirty global targets (including large tech firms, financial institutions, chemical manufacturers, and government agencies), achieving some successful compromises; the report highlights this as the first large-scale AI-driven cyberattack and emphasizes the roles of advanced model capabilities, agentic autonomy, and tool access in enabling the operation.