Anthropic reported Thursday that Chinese hackers utilized its AI chatbot, Claude, in what it considers the first cyberespionage operation predominantly executed through artificial intelligence.
The company stated cybercriminals employed Claude to target approximately 30 technology companies, financial institutions, chemical manufacturers, and government agencies. Hackers used the AI platform to collect usernames and passwords from company databases, subsequently exploiting these credentials to steal private data. A “small number” of these attacks succeeded, Anthropic noted.
“We believe this is the first documented case of a large-scale cyberattack executed without substantial human intervention,” Anthropic said.
The Wall Street Journal first reported this development. Anthropic detected suspicious activity beginning in mid-September. An investigation linked the activity to an espionage campaign likely conducted by a state-sponsored Chinese group. The investigation indicated hackers deceived Claude into believing it was a legitimate cybersecurity firm employee performing defensive tests. Cybercriminals also broke down the attack into small tasks to obscure their actions.
The operation required minimal human intervention. “The AI made thousands of requests per second, an attack speed that would have been, for human hackers, simply impossible to match,” Anthropic stated.
Anthropic anticipates a rise in the scale and sophistication of AI cyberattacks as AI agents become more prevalent.
