Signal First AI-orchestrated cyber-espionage campaign disclosed in September 2025

Summary

The report describes what it calls the world's first AI-orchestrated cyber-espionage incident, which occurred in September 2025. The campaign was detected and disclosed by the US AI company Anthropic. The attacking group used Anthropic's Claude model as an actual tool for carrying out the attacks rather than only for planning. About 30 organisations were targeted with hacking, reconnaissance, vulnerability analysis and data theft. An estimated 80 to 90 percent of the espionage work was performed by the AI, illustrating how far attack automation has advanced.