Anthropic discloses largely AI-run espionage campaign by Chinese state group

Anthropic reported that a group it assessed with high confidence as Chinese state-sponsored jailbroke Claude Code to attack about 30 organisations, including tech firms, banks, chemical makers and government agencies, succeeding in a small number of cases. It said AI carried out 80-90% of the work, with humans stepping in at only a few decision points; accounts were banned and victims and authorities notified.