• Wion
  • /Trending
  • /AI chatbots prepare detailed terrorist attack plan, suggest which places to bomb in US...all in 12 hours

AI chatbots prepare detailed terrorist attack plan, suggest which places to bomb in US...all in 12 hours

AI chatbots prepare detailed terrorist attack plan, suggest which places to bomb in US...all in 12 hours

ChatGPT, Gemini, Grok and Claude told investigators how to attack the US. Photograph: (AI generated image)

Story highlights

Four of the biggest AI platforms prepared a detailed terror attack plan, suggesting where to place the bombs, specific targets, hitting electrical grids, with one of them even proposing the use of a drone to make things easier, all within 12 hours. 

Artificial Intelligence is all set to wreak destruction across the world, as a US report on Wednesday revealed that four of the major chatbots helped them plot a terror attack and even told them where to place the bombs. According to the Government Accountability Office report, the officials were able to break through ChatGPT, Claude, Gemini and Grok despite all four of them having safeguards to prevent such misuse. This is a scary scenario considering all the recent warnings and threats around AI.

"We were four for four. The safeguards did not work," Kevin Walsh, a director of IT and cybersecurity audits at the GAO, told Hearst Television. Each of the chatbots created a detailed plan for the attack in less than 12 hours, suggesting targets, locations, timing and methods. Two of them helped plan the bombings, while the other two detailed attacks on America's power grid, with one of them even suggesting using a drone.

One of the platforms presented a detailed 15-step plan to execute the terror attacks. The report has caused a stir about the many ways bad players can exploit AI, including a large-scale terror attack. The investigators who experimented said they were able to easily bypass the safeguards, and warned that the bots “produced feasible information that could assist a malicious actor with planning and executing attacks.”

Experts said the terrorist plan will cause explosions


"Specifically, in response to prompts, two platforms produced information for constructing and detonating an improvised explosive device (IED) designed to cause mass casualties," GAO stated in the report. One of the bots presented a 15-step plan. The investigators shared the plans with experts to see how legit they were.

Trending Stories

A bomb squad commander was also shown the bombing plans the four AI platforms made. All of them confirmed that if the instructions were deployed, they would produce explosions. "We assess that these were viable attack plans. And even if, you know, we didn't go through with it, we felt that these could have helped someone who set out to do evil," Walsh said.

Senator Maggie Hassan, incoming chair of the Senate Homeland Security and Governmental Affairs Committee, said that she will plan hearings on AI safety next year and called for Congress to act quickly. "One of the things that will be necessary going forward to make sure that they're all dealing with a level playing field when it comes to safety is making sure that before models are released, they have safeguards that can't be breached in this way," she said.

About the Author

Share on twitter

Anamica Singh

Anamica Singh is a Senior News Editor at WION, bringing over 17 years of deep media and journalism experience to the platform. Specialising in high-impact global journalism, she le...Read More