AI Company Claims Chinese Hackers Used Its Technology for Automated Cyber Attacks

AI developer Anthropic has claimed that Chinese state-sponsored hackers exploited its chatbot Claude to conduct automated cyber attacks against around 30 global organizations. According to the company, the hackers tricked Claude into performing automated tasks under the guise of cybersecurity research, in what Anthropic describes as the “first reported AI-orchestrated cyber espionage campaign.”

Researchers say the attacks were detected in mid-September, and the hackers used small automated tasks strung together into a sophisticated espionage operation. Targets included major tech companies, financial institutions, chemical manufacturers, and government agencies. With Claude’s coding assistance, the hackers were allegedly able to autonomously breach targets, extract sensitive data, and analyze it for valuable insights.

Despite the sensational claims, experts remain skeptical. Critics argue that AI is still too unreliable to carry out fully autonomous cyberattacks. Anthropic admitted that Claude made errors, such as generating fake login credentials and misreporting publicly available information as secret.

The company emphasizes that the same capabilities that allowed the AI to be exploited could also make it an essential tool for cyber defense. AI defenders could be the key to stopping other AI-powered threats, Anthropic claims.