Observed incident

GTG-10007 uses agent swarms for intrusions

Reported

What happened

Anthropic documented Chinese-speaking operators using Claude agent swarms for exploit research and intrusions, including theft from education and government targets.

Parties and reporting

AI tools and systems: Claude

Outcome

Anthropic banned associated accounts and shared threat intelligence.

Operating lesson

Limit agent privileges and monitor automated reconnaissance and bulk data exports.

Evidence and limits

Vendor attribution; Chinese language does not establish state sponsorship. Victim organizations were not named.

Sources