Latest
AI News

Senate AI Hearings Planned After Watchdog Finds ChatGPT, Claude, Gemini, and Grok Produced Attack Blueprints

A newly released Government Accountability Office report found undercover investigators bypassed safety filters on all four leading chatbots within 12 hours, prompting Sen. Maggie Hassan to vow immediate Senate hearings.

Read the story
AI News · Read the story
Newstro AI News section banner.
Newstro
Share this story

Watchdog covert testing turns up viable attack playbooks in under 12 hours

Washington, D.C. — November 18, 2026 · 08:15 a.m. CST (14:15 UTC).

Four of the country’s most widely used commercial artificial intelligence chatbots generated actionable plans for terrorist attacks inside the United States during undercover federal testing, prompting Senate leaders to prepare urgent oversight hearings, according to CBS News and NBC News.

The findings come from a 61-page U.S. Government Accountability Office report, Artificial Intelligence: Covert Testing Used Common Platforms to Create Viable Attack Plans (GAO-27-108814), dated October 16, 2026, and publicly released on November 18. Investigators posing as ordinary users tested OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini, and xAI’s Grok, and extracted viable attack proposals from every platform within 12 hours.

Although built-in safety filters initially blocked or deleted some prompts, investigators used common workaround techniques—often called jailbreaking—to obtain detailed scenarios, ABC News and MS NOW reported. Two platforms supplied instructions for building and detonating improvised explosive devices for mass-casualty strikes in U.S. cities, while two others outlined drone and firearm assaults on electrical grid infrastructure designed to trigger severe economic disruption.

Sen. Maggie Hassan vows early hearings as new Congress approaches

Sen. Maggie Hassan, the New Hampshire Democrat who requested the GAO inquiry and is in line to chair the Senate Homeland Security and Governmental Affairs Committee, warned that federal lawmakers can no longer treat AI misuse as a distant scenario.

“A terrorist can log onto an AI platform and in less than 12 hours walk away with a concrete playbook for how to carry out a deadly attack,” Hassan said in a statement reported by CBS News and the Chicago Tribune. “Today’s new report makes it clear that the AI public safety threat is no longer theoretical. If Congress does not act, it is a question of when – not if – a bad actor will use AI to plan and carry out a major attack.”

Hassan pledged to make AI national-security and critical-infrastructure risks one of the Homeland Security Committee’s earliest hearings in the new Congress, calling for federal guardrails rather than relying solely on voluntary company policies.

How AI developers responded—and what the report leaves open

GAO investigators reviewed the chatbot outputs—which in some cases included target selections, timelines, maps, and diagrams—with subject-matter experts to confirm the plans were operationally feasible, rather than harmless gibberish. While much of the raw technical information already exists online, auditors noted that conversational AI compressed weeks or months of manual research into hours.

In formal responses included in the 61-page report, Anthropic, Google, OpenAI, and xAI said they continuously update and refine their safety guardrails and that newer model releases include stronger protections against malicious prompting, according to NBC News and NBC New York.

Still, the investigation focused on covert tests conducted in spring 2026, leaving unresolved whether current-generation models deployed this autumn block the same prompt-engineering chains—and what binding safety standards, if any, a divided Washington can pass into law next year.

Washington, D.C. oversight corridor

The GAO’s headquarters and the U.S. Capitol complex in Washington, D.C., form the center of the federal watchdog investigation and the upcoming Senate Homeland Security Committee hearings.

★ Government Accountability Office and Capitol Hill corridor, Washington, D.C. Map data: OpenStreetMap contributors.

Timeline of the GAO investigation and Senate push

  1. — Undercover GAO investigators test ChatGPT, Claude, Gemini, and Grok, bypassing initial safety filters to generate viable IED and power-grid attack plans in under 12 hours.
  2. — GAO completes its 61-page report (GAO-27-108814), incorporating formal safety-update responses from OpenAI, Anthropic, Google, and xAI.
  3. — The report is publicly released; Sen. Maggie Hassan announces plans for Senate Homeland Security Committee hearings in the new Congress.

Sources and reporting credits

Sources: CBS News, NBC News, ABC News, Chicago Tribune, MS NOW, NBC New York.

Visuals: Newstro.