Anthropic reports Claude use in missile guidance and espionage operations
The AI company says it disrupted alleged weapons, cyber-espionage, influence and surveillance operations, but found no evidence a working missile was deployed.
Anthropic says its Claude AI models were used in alleged missile-guidance work in northern Yemen, cyber-espionage against Ukrainian, European and diplomatic organisations, and other influence and surveillance operations.
The company alleges operators assigned separate Claude instances to develop guidance and flight-control software for a guided rocket and a long-range ballistic missile. Anthropic says internal safeguards blocked many requests, but some activity escaped detection after the operators divided tasks across separate sessions.
Anthropic says it has no evidence that a working weapon was deployed and that an apparent test-fire was unsuccessful. The company says it banned the associated accounts and shared information with public- and private-sector partners.
The report also alleges that a Russian-linked operation bearing the hallmarks of Midnight Blizzard, or APT29, used automated AI workflows for phishing, system set-up and data theft targeting Ukraine, Europe and diplomatic organisations, including drone makers. Anthropic says it disrupted the operation and added monitoring to identify similar activity.
Separately, Anthropic says a Chinese operation involving university students in Hunan province used Claude in an offensive cyber programme targeting government and corporate networks across the Middle East, Europe and Southeast Asia. The company also alleges that Iranian state-aligned accounts used the models for covert influence and psychological operations, while China-aligned operators used Claude to profile targets and conduct a recruitment effort aimed at Uyghurs in Syria.
The supplied material does not independently establish the identities or state links of the alleged operators, or how much of the activity was directed by people rather than conducted autonomously. The report comes amid continuing tensions between Anthropic and the Pentagon over safeguards for autonomous weapons and domestic surveillance.
