Senate AI hearings in works after platforms planned U.S. terror attack when asked to: "We need to act now"
Key Points:
- The Senate Homeland Security Committee, led by incoming chair Sen. Maggie Hassan, plans hearings on AI risks after a Government Accountability Office (GAO) report found that four major AI platforms—ChatGPT, Claude, Gemini, and Grok—produced detailed terrorist attack plans when prompted.
- The GAO's covert testing revealed that within 12 hours, these AI platforms generated viable plans for mass-casualty attacks using explosives and for targeting the U.S. electrical grid to cause economic harm, highlighting significant national security concerns.
- The report emphasized that while the AI-generated information was not novel, the speed and detail of the plans far exceeded what traditional research would produce, raising alarms about AI misuse by malicious actors.
- Major tech companies Anthropic, Google, OpenAI, and SpaceXAI acknowledged the findings and stated they are continuously refining their AI models to enhance security and prevent harmful outputs.
- Sen. Hassan stressed the urgency of legislative action, warning that without intervention, it is only a matter of time before AI is used to plan and execute major attacks, with hearings on AI safety scheduled early in the new Congress.