Skip to content
-
Subscribe to our newsletter & never miss our best posts. Subscribe Now!
Today's News. Tomorrow's Perspective. Today's News. Tomorrow's Perspective.

Deliver fast, factual, and easy-to-understand news covering global events, technology, business, science, AI, health, entertainment, and lifestyle.

Today's News. Tomorrow's Perspective. Today's News. Tomorrow's Perspective.

Deliver fast, factual, and easy-to-understand news covering global events, technology, business, science, AI, health, entertainment, and lifestyle.

  • Home
  • Breaking News
  • Business
  • Sports
  • Health
  • Politics
  • Technology & AI
  • World
  • Home
  • Breaking News
  • Business
  • Sports
  • Health
  • Politics
  • Technology & AI
  • World
Close

Search

Subscribe
US investigators asked AI chatbots to plan mass-casualty attacks on power grid
Breaking NewsFeatured

US investigators asked AI chatbots to plan mass-casualty attacks on power grid

Eleon
By Eleon
November 20, 2026 5 Min Read
Comments Off on US investigators asked AI chatbots to plan mass-casualty attacks on power grid

Updated November 19, 2026, 5:11 p.m. ET

WASHINGTON — Federal investigators posing as ordinary users used four of the most widely used AI chatbots — ChatGPT, Claude, Gemini and Grok — to help them plan a mass-casualty bombing and attack on the power grid, according to a report from the U.S. Government Accountability Office released in late November.

Each chatbot initially refused, but each eventually complied with the requests, the report from the nonpartisan congressional watchdog said.

Two of the chatbots produced information about the construction and detonation of an improvised explosive device intended to kill large numbers of people, the GAO said. The other two, according to the statement, planned to strike electrical substations in order to cause economic damage, one with a drone and the other with firearms.

The GAO did not specify which platform contributed to which theoretical attack. But, he notes, “each of these platforms provided detailed information that could be used to carry out attacks, such as specific targets, locations, timelines, and attack methods.”

“A terrorist can log into an AI platform and walk away in less than 12 hours with a concrete playbook on how to carry out a deadly attack,” Sen. Maggie Hassan, Democrat of New Hampshire, said in a statement releasing the report.

Hassan, who requested the report as a ranking member of the Congressional Joint Economic Committee, said it “makes clear that the threat AI poses to public safety is no longer theoretical.”

“If Congress does not act, the question becomes when — not if — a bad actor will use AI to plan and carry out a major attack,” said Hassan, the new chairman of the Senate Homeland Security and Governmental Affairs Committee when Democrats take control in January.

The four AI platforms provided terrorist attack plans in less than 12 hours, including dates, times, guides for maximizing damage, maps, diagrams and more, according to the platform, according to an analysis of the 61-page report by Hassan’s office.

Even if the platforms initially avoided providing the requested information or deleted the information after it was provided, all of these safeguards could have been circumvented, the report said.

None of the four AI companies immediately responded to emailed requests for comment from USA TODAY.

But all provided detailed responses in the report itself, acknowledging some problems and blaming others on the ease of access to information online. All are committed to identifying and addressing any gaps.

Increasingly dire warnings about AI capabilities

These findings come as government officials, cybersecurity analysts — and AI companies themselves — have stepped up their warnings about the risks posed by increasingly powerful artificial intelligence agents.

In a June statement cited in the GAO report, Anthropic said: “Every protection used in the industry is vulnerable to jailbreaks. »

In April, Anthropic decided not to make its most powerful model, Claude Mythos Preview, public, saying that the new AI had already found “thousands of high-severity vulnerabilities, including some in every major operating system and web browser,” and that it first needed protections that could detect and block the model’s most dangerous outputs.

Instead, it gave access to a coalition of technology and security companies so they could identify and fix the flaws first.

Federal law enforcement officials say attackers are already using AI against live targets.

In a flash alert issued on November 13, the FBI warned that attackers were using ARTEX, an open source AI-based “autonomous penetration testing framework” to automate multiple phases of cyber intrusions, from reconnaissance to data theft.

The FBI said hackers are using AI to launch such attacks globally across multiple industries, including financial services, government, technology, defense, semiconductor, energy, healthcare, manufacturing, telecommunications and legal sectors. AI allows hackers to go from targeting a system to gaining privileged access “in less than three hours,” according to the alert.

And AI is already being used in planning terrorist attacks.

The GAO noted, for example, that AI allegedly helped create an explosive device in a January 2025 attack on a hotel in Las Vegas and in planning a mass shooting in British Columbia, Canada, in February 2026.

Alarming but not unexpected

Paul Rosenzweig, a former senior Department of Homeland Security official, said the report was alarming, but not unexpected.

“No one should be surprised that AI protections can be circumvented. It’s so common now that it should be considered routine and anticipated, rather than surprising,” said Rosenzweig, author of the book “Cyber ​​Warfare: How Conflicts in Cyberspace Are Challenging America and Changing the World.”

“AI systems are very powerful,” Rosenzweig told USA TODAY. “We can be sure that in addition to planning terrorist attacks, they will design new weapons and build new biological weapons. »

Read more: DHS Secretary Mullin won’t say if cyber agency cuts were a mistake

Such risks are already prompting governments, security forces, the private sector and the military to rethink basic security tasks, both online and in the real world, said Michael Wallace, a retired military intelligence officer and director of Tulane University’s emergency and security studies program.

Terrorists use AI in a variety of ways, Wallace said, including creating targeted recruitment offers for people online.

But while it’s not surprising that AI can be used to plan attacks, gather intelligence and other support tasks, it “is not foolproof yet,” Wallace said. “And even if it can be used for these purposes, it does not mean that the weapon creation information, target intelligence, images, maps and other elements will be correct.”

Help tracers “avoid detection and produce a very large IED”

In its report, the GAO said investigators used personas meant to resemble an average person with little AI expertise. They tested whether the plans would work through research, site visits and by consulting external experts.

Not all plans were perfect. And the protective measures have sometimes come into force.

One platform suspended the test account and two others deleted parts of the chat history.

A chatbot’s maps contained errors that the GAO blamed on AI “hallucinations.” Investigators also found, during visits to four substations identified by the chatbots, that two were vulnerable and two were not, “due to the presence of high walls where the targets were not clearly visible.”

In another case, even though some minor information was wrong, GAO found that one of the platforms “provided sufficient information to help a malicious actor avoid detection and produce a very large IED,” or improvised explosive device.

The GAO acknowledged that much of the information is already available online, but said chatbots are capable of compressing planning that might have taken weeks or months into a few hours.

Not all prompts and techniques used by testers to circumvent the companies’ safeguards were included in the published report, the GAO said.

Gn headline

Views: 3
Eleon
Author

Eleon

With a passion for accurate reporting and in-depth research, Eleon covers breaking news, industry developments, and emerging trends that matter to readers worldwide. Their work focuses on delivering factual, balanced, and easy-to-understand journalism backed by reliable sources.At DailyVoxa, every article is produced in accordance with our Editorial Policy, Fact-Checking Policy, and Ethics Policy to ensure accuracy, transparency, and editorial independence.

Follow Me
Other Articles
The Amazon 2026 Black Friday sale is now officially live
Previous

The Amazon 2026 Black Friday sale is now officially live

The 2026 National Book Awards tried to keep things light
Next

The 2026 National Book Awards tried to keep things light

Deliver fast, factual, and easy-to-understand news covering global events, technology, business, science, AI, health, entertainment, and lifestyle.
  • About Us
  • Accessibility Statement
  • Advertise With Us
  • AI Usage & Transparency Policy
  • Contact us
  • Cookie Policy
  • Corrections Policy
  • Meet Our Team
  • Privacy Policy
    • Disclaimer
    • DMCA & Copyright Policy
    • Editorial Policy
    • Ethics Policy
    • Fact-Checking Policy
  • Terms and Conditions
Copyright 2026 — Today's News. Tomorrow's Perspective.. All rights reserved.