Government Watchdog Uncovers AI Platforms Generating Terrorist Attack Plans

Washington:Federal investigators have revealed that major artificial intelligence platforms can produce detailed plans for terrorist attacks, bypassing existing safety safeguards. This finding comes from a government watchdog report released on Wednesday, highlighting vulnerabilities in AI technologies.

According to TourismAfrica2006, the Government Accountability Office (GAO) conducted undercover testing in the spring of 2026 on platforms including ChatGPT, Claude, Gemini, and Grok. The investigation discovered that these platforms generated information to assist in attacks on crowded locations and the nation's electrical grid, all within less than 12 hours.

The GAO report indicated that two AI platforms devised plans involving improvised explosive devices aimed at causing mass casualties, while the other two outlined attacks on electrical infrastructure to inflict economic damage. Despite initial activation of content removal safeguards by all platforms, investigators managed to circumvent these protections, obtaining feasible attack plans.

Experts consulted by GAO verified the feasibility of the AI-generated plans, raising concerns about the capability of AI to aid malicious actors. The companies behind these platforms—OpenAI, Anthropic, Google, and SpaceXAI—have expressed their commitment to enhancing safety measures. However, Anthropic and Google highlighted challenges and limitations in the testing methodology and model versions used during the investigation.

The report, requested by Senator Maggie Hassan, underscores the urgency of addressing AI's potential misuse, as she warns of the real threat AI poses to public safety. Hassan emphasized the necessity for congressional action to prevent AI from being used in executing major attacks.