Anthropic, Claude and hackers
Digest more
Anthropic said GTG-1002 developed an autonomous attack framework that used Claude as an orchestration mechanism that largely eliminated the need for human involvement. This orchestration system broke complex multi-stage attacks into smaller technical tasks such as vulnerability scanning, credential validation, data extraction, and lateral movement.
The attack involved hackers tricking Anthropic's AI tool, Claude Code. Around 30 companies, institutions, and agencies were targeted.
Chinese state-sponsored attackers used Anthropic’s Claude AI to automate 90% of an advanced cyber espionage campaign, breaching four organizations in just 48 hours. By breaking attacks into innocent-looking tasks,
Anthropic announced on Thursday that Chinese state-backed hackers used the company’s AI model Claude to automate roughly 30 attacks on corporations and governments during a September campaign, according to reporting from the Wall Street Journal.