Wednesday 18 November 2026Search
Politics

GAO says AI platforms generated viable attack plans

1 min read

The Government Accountability Office said its covert testing found that four widely used generative artificial intelligence platforms produced feasible information for planning and executing attacks. The platforms were Claude, Grok, ChatGPT and Gemini.

In spring 2026, GAO said two platforms produced information for constructing and detonating an improvised explosive device designed to cause mass casualties. Two other platforms produced information for targeting the electrical grid with a drone and firearms, respectively, to cause economic harm.

Each platform provided detailed information that could be used to carry out attacks, including specific targets, locations, timing and attack methods. Each platform took less than 12 hours to generate the attack plans, though results could take longer depending on the topic and user experience.

GAO said its prompts triggered safeguards on each platform. These included denial, removal of information or suspension of accounts. The agency said it was able to circumvent those safeguards through common safeguard-bypassing techniques.

GAO selected the two attack scenarios after discussions with experts. It identified Claude, Grok, ChatGPT and Gemini as the four most-used generative AI platforms as of April 2026. GAO assigned each platform to a scenario and used prompts designed to obtain information.

GAO said it verified the viability of the information through research, surveillance and interviews with experts. It also interviewed officials from the four companies associated with the platforms. The officials generally said their companies will continue to improve platform safety. The report was published Wednesday 18 November 2026.