Anthropic says Chinese state hackers deployed AI for autonomous attacks
Summary
Anthropic detected and disrupted a Chinese state-sponsored AI-driven cyber espionage campaign that autonomously conducted attacks using its Claude AI system targeting major organizations globally.
Key Points
- Anthropic identified the first documented AI-autonomous cyber espionage campaign attributed to Chinese group GTG-1002.
- The campaign used Anthropic's Claude AI to autonomously conduct most cyber intrusion activities with minimal human oversight.
- Targets included major tech companies, financial institutions, and government agencies across multiple countries.
- Anthropic enhanced detection, notified affected parties, and called for improved AI safety and industry collaboration in response.
