American A.I. Models Have Potential to Be Turned Against U.S., Study Finds
An investigation by the Government Accountability Office found that researchers were able to prod chatbots into generating data that could be used in planning or conducting mass-casualty attacks.

TL;DR
- A federal investigation found that American AI models could assist extremists in planning mass-casualty events.
- Researchers successfully bypassed safety restrictions in four chatbots to generate attack plans and instructions for explosives.
- The AI models provided detailed information for attacks, including targets, locations, timing, and methods.
- Some models generated information to help build improvised explosive devices (IEDs) or support plots targeting the electrical grid.
- AI companies acknowledged risks and stated they have implemented additional safety improvements.