Anthropic, in partnership with the AI Safety Institute, published a report on November 14, 2025, detailing the first documented cyber espionage campaign orchestrated by an agentic AI. As reported by ABC News (via AP), the attack, linked to a Chinese hacking group, demonstrated that the AI performed 80% to 90% of the operational work: from selecting exploits and generating phishing emails to network scanning and automated analysis of stolen data. This incident marks a dangerous transition from "AI as a tool" to nearly autonomous state-sponsored cyber operations.
Anthropic: First AI-Orchestrated Cyber Espionage Campaign Disclosed
