Senate hearings on AI underway after platforms planned US terror attack when asked: ‘We must act now’
Washington — The new chairman of the Senate Homeland Security Committee is preparing hearings on AI after government watchdogs found that the four most widely used platforms – ChatGPT, Claude, Gemini and Grok – had all created proposals for a terrorist attack in the United States at the request of investigators.
Democratic Sen. Maggie Hassan, who will lead the panel in the new year, has tasked the Government Accountability Office with investigating how commercially available AI could be used for malicious purposes, posing national security risks.
According to a report released Wednesday, the GAO found that within 12 hours of its secret tests, the platforms were able to create viable plans to use explosive devices for mass-casualty attacks in U.S. cities or to attack parts of the nation’s power grid to create economic damage.
“We need to act now,” Hassan said in a conference call with reporters.
The senator announced that one of the first hearings she will hold when the new Congress takes office in January will focus on the risks of AI and ways to improve security.
“The threat AI poses to public safety is no longer theoretical,” Hassan said in a statement. “If Congress fails to act, the question becomes when – not if – a bad actor will use AI to plan and carry out a major attack.”
The 61-page GAO report includes responses from tech companies — Anthropic, Google, OpenAI and SpaceXAI — who have largely said they continue to refine their platforms to ensure security.
“The details behind the information provided by AI platforms were not new,” the GAO wrote. “Many details are available on the Internet. However, the generation of plans that could have taken weeks or months have been achieved in a few hours with the help of AI. These platforms could also spark ideas or methods that would not have been easily revealed by basic research.”
“The potential benefits of this technology are immense, but they can also benefit malicious actors,” the GAO added.
“As AI capabilities evolve, AI vendors will continue to refine their protective measures,” GAO said. “However, in its current state, this technology can be potentially dangerous and its misuse represents a growing risk to national and economic security.”
Plans developed by AI platforms
GAO investigators conducted the performance audit from December 2025 through October 2026, including testing in spring 2026.
In response to requests, two of the AI platforms produced plans to construct and detonate an improvised explosive device (IED) to cause mass casualties, while two other platforms produced plans to use drones or firearms to target the power grid, according to the report.
“Each of these platforms provided detailed information that could be used to conduct attacks, such as specific targets, locations, timing, and methods of attack,” the GAO found. “Each platform took less than 12 hours to generate the attack plans.”
In one of the tests, the platform suggested renting a vehicle and picking up supplies from a store along a route to the location. He recommended a specific date and time for the attack, the report said, because public events likely to bring together large crowds were scheduled.
While the GAO said the rig initially refused to respond to requests for detailed information on how to build and detonate the explosive device, it said it managed to get around its guardrails and get the rig to provide the information.
The other tests had similar results.
For tests involving attacks on electrical substations, one of the platforms suggested a drone attack while the other suggested the use of firearms to attack substations near a U.S. airport – both designed to inflict economic damage. The platforms presented maps and images.
GAO investigators consulted with experts and conducted site visits to verify the accuracy of the plans provided by the AI platforms.
In one case, they found that roads were poorly placed, and in another case, they noted the presence of high walls that would make it difficult to see targets.
AI companies say they are updating their models
In their responses to the GAO report, big tech companies largely said their products continue to be refined.
Anthropic, in a letter to the GAO included in the report, said it does not dispute that a determined tester could extract harmful content from an AI system by misrepresenting the purpose of a request, “and we take this conclusion seriously.” But he said he was unable to verify the GAO report’s findings.
Google said GAO’s testing was done on previous models and that “we continue to improve our models.”
OpenAI said it welcomed the opportunity to respond to the GAO’s findings, “as well as the work that remains.”
SpaceXAI said it “continues to work diligently” to improve safety with each model release.
Gn headline