Incident·WS-0089·Mar 2026·disclosed Sep 10, 2026 (179 days later)·in the wild·AI-powered cyberattacks

Lone hacktivist uses Claude against 42 European political groups

Anthropic Claude; actor GTG-50029 (single French-speaking hacktivist)

56/100
WS-3 Serious
Harm score 65, signal score 46. Autonomy: acted (the AI took a bounded action). How the scores work.
Harm 3/5
damage actually done
Reach 3/5
who was affected
Autonomy 2/4
how much the AI did on its own
Novelty 2/5
how new the behavior was
Permanence 4/5
how hard it is to undo

Anthropic's September 2026 threat report says a lone hacktivist used Claude in spring 2026 to build attack tooling. The targets were European political parties, media outlets, think tanks and their service providers. The actor got inside at least 14 of 42 tracked organizations, leaked data and set up doxxing lookups. Anthropic banned the accounts.

Damage

~140,000 records (incl. political opinions) taken from a campaign-management platform; 12–26 GB of database dumps overall; at least 14 organizations accessed

Sources

Anthropic. Evidence confidence: medium. This is the only primary source; victims are not named. Infrastructure was active 6 Feb–4 Jul 2026, so some activity may fall outside the window.

Factor ratings and the score are this index’s judgments by a published formula; facts, dates and figures come from the linked sources. Where a lawsuit is pending, the summary reports allegations, not findings.