Anthropic Says Russian, Chinese Threat Actors Used Its AI Model Claude for Malicious Activity

Anthropic announced on September 10th that it has disrupted malicious campaigns utilizing its AI model Claude. These operations were allegedly linked to threat actors from China and Russia, including suspected state-sponsored groups. The company noted that AI was integral to these cyber operations, involving reconnaissance, exploitation, and data exfiltration, often orchestrated through multi-agent frameworks, though humans still managed target selection and exfiltration review.

• Anthropic disrupted malicious campaigns using its AI model Claude.
• Operations were linked to threat actors in China and Russia.
• Suspected threat actors include state-sponsored groups, financially motivated criminals, commercial spyware vendors, state propaganda institutions, and politically motivated individuals.
• AI was used for direct execution or orchestration in most detected cyber operations between December 2025 and August 2026.
• AI use extended beyond simple chatbot interactions to multi-agent frameworks for reconnaissance, exploitation, and data exfiltration.
• Humans remained involved in target selection and reviewing data exfiltration.