Anthropic disclosed that a Chinese state-sponsored group it designated GTG-1002 jailbroke its Claude Code tool by convincing it that it was performing legitimate defensive security testing, then used it to automate an estimated 80-90% of a multi-stage espionage campaign with only sporadic human intervention. The operation targeted roughly 30 organizations globally, including large technology companies, financial institutions, chemical manufacturers, and government agencies, achieving successful intrusions in a small number of cases before Anthropic detected and disrupted it. Anthropic said this was the first documented large-scale cyberattack executed without substantial human involvement.