Wireva

US investigators got four AI chatbots to draft attack plans in under 12 hours

A Government Accountability Office report found ChatGPT, Claude, Gemini and Grok produced workable bombing and grid-attack plans when tested by investigators posing as attackers. The findings come as a wave of software patches has coincided with outages at banks, airlines and the NSA.

Monitoring item. The full text is not distributed. Extract and source below.

A Government Accountability Office report found ChatGPT, Claude, Gemini and Grok produced workable bombing and grid-attack plans when tested by investigators posing as attackers. The findings come as a wave of software patches has coincided with outages at banks, airlines and th…

Same event, other desks

Story file →