Government investigators find AI platforms produce terror plot plans in undercover op

Read the original at Fox News ↗
Cite

Sorace, S. (2026, November 18). Government investigators find AI platforms produce terror plot plans in undercover op. Fox News. https://www.foxnews.com/politics/government-investigators-find-ai-platforms-produce-terror-plot-plans-undercover-op

Dated from the outlet's own feed. Check the original for its byline.

Fox News · collected 2026-11-18 · by Stephen Sorace

Quick Summary

Government investigators bypassed safety measures on four AI platforms—ChatGPT, Claude, Gemini, and Grok—in an undercover operation and obtained plans for terrorist attacks within 12 hours. The Government Accountability Office (GAO) report, released Wednesday, details how these platforms provided information on potential targets, attack methods, and timing for deadly and destructive acts. Despite initial safety safeguards, the investigators circumvented these protections, and expert reviews confirmed the feasibility of the plans provided. The companies behind the platforms acknowledged the need for improvements in safety measures but questioned some aspects of the GAO's testing methods.
Written by AI on 2026-11-21, from this article's own text only.

AI analysis

Why this leaning score
The model judged this article politically coded and scored it +0.35, but all 2 of its quote(s) are attributed speech - words the article quotes from someone, not the article's own narration, so the score is not published.
Written under an earlier scoring contract, which gave a paragraph rather than checkable quotes. Re-analysing this article replaces it.
Leaning score withheld for article 163798: attributed speech only · logged 2026-11-21

Compared with similar articles

Nothing to compare against. No article is close enough to this one for the pipeline to have linked or judged the pair.
How this article was measured signals, leaning charts, publisher record, byline, topics and claims

Signals How these are calculated →

Claims extracted
14
sentences that state something as happening or true
Uncertain
21%
3 of 14 hedged ("reportedly", "may", "according to")
Leaning
withheld
no quote in the article backed the model's score
Publisher signals
66/100
Fox News's corrections and hedging, across everything collected from it. Not a measure of accuracy.
Outlets on this story
unclustered
not grouped into a story yet
Same framing elsewhere
1
articles that frame the event the way this one does
Analyzed 2026-11-21 · how these are computed

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Publisher

Fox News · 8391 article(s) · 16 correction(s) detected
Running correction rate · 16 correction(s)

Who wrote this

Stephen Sorace
35 article(s) here · 1 carrying a prediction
The Government Accountability Office (GAO) found that ChatGPT, Claude, Gemini and Grok each generated information that could help carry out attacks targeting crowded locations or the nation's electrical grid during undercover testing conducted in spring 2026.
Duffy said President Donald Trump convened federal officials on Friday as the administration investigated whether the disruption could have involved malicious interference.
"There will be no more," Aubrey Vaughan told FOX 5 DC, referring to the party for their sons.
In Texas, John Allen Rubio, 46, was set to be executed that evening for the 2003 stabbing and beheading of three young children he was raising in Brownsville.
In a Truth Social post, Trump singled out Minnesota, California and Illinois as examples of states he claims are plagued by corruption and mismanagement, arguing that cracking down on the alleged criminal activity could help balance the federal budget.
The congresswoman said she could see the Capitol from her vehicle and instructed her staff to notify House officials that she was delayed.
The two survivors were expected to be transported to an area hospital, though the extent of their injuries was not immediately known.
Trump, speaking to Fox News Channel during live Election Night coverage, said he was surprised by the GOP's performance but expressed confidence that he could work with Democrats despite their electoral gains.
Voters in line when polls close will be permitted to cast their ballots.
We may find parts of the aircraft some distance away from the actual impact site."
Also by Stephen Sorace
Nothing else under this byline is closely related to this article, so these are simply their most recent.
All 35 articles by Stephen Sorace →

Topics

Anthropic GAO Google OpenAI SpaceXAI

Subjects

GAO ORG · 4× Anthropic ORG · 2× Google ORG · 2× OpenAI ORG · 2× SpaceXAI ORG · 2× ChatGPT ORG · 1× Hassan PERSON · 1× Maggie Hassan PERSON · 1× N.H. GPE · 1× The Government Accountability Office ORG · 1×

Narrative

GAO investigators also consulted experts and conducted real-world checks to assess the plans, finding that the information provided by all four platforms was sufficiently feasible to assist someone planning an attack.
framing: mixed · carried by 1 article(s) · first seen 2026-11-21
The Government Accountability Office (GAO) found that ChatGPT, Claude, Gemini and Grok each generated information that could help carry out attacks targeting crowded locations or the nation's electrical grid during undercover testing conducted in spring 2026.

Claims (14 extracted, 3 hedged)

Federal investigators successfully bypassed safety safeguards on four major artificial intelligence platforms, obtaining detailed plans for potentially deadly terrorist attacks in less than 12 hours, according to a government watchdog report released Wednesday. uncertain
investigators → bypass → report
The Government Accountability Office (GAO) found that ChatGPT, Claude, Gemini and Grok each generated information that could help carry out attacks targeting crowded locations or the nation's electrical grid during undercover testing conducted in spring 2026. uncertain
that → find → spring
Two platforms produced plans involving improvised explosive devices intended to cause mass casualties, while the other two outlined attacks on electrical infrastructure designed to inflict economic damage. asserted
two → produce → damage
The AI-generated responses included potential targets, locations, timing and attack methods. asserted
responses → generate → targets
Although all four platforms initially activated safeguards to block or remove harmful content, investigators ultimately circumvented those protections. asserted
investigators → activate → protections
GAO investigators also consulted experts and conducted real-world checks to assess the plans, finding that the information provided by all four platforms was sufficiently feasible to assist someone planning an attack. asserted
information → consult → attack
"The potential benefits of this technology are immense, but they can also benefit malicious actors," the GAO warned, citing growing risks to national and economic security. asserted
GAO → benefit → security
The companies behind the platforms — OpenAI, Anthropic, Google and SpaceXAI — said they were continuing to improve their safety measures, while some raised concerns about the testing. asserted
some → say → testing
Anthropic acknowledged that determined users could bypass AI safeguards and said it took the findings seriously, but questioned aspects of the GAO's methodology and said it could not independently verify the results. uncertain
it → acknowledge → results
Google noted that investigators tested earlier versions of its models, saying, "We continue to improve our models." asserted
We → note → models
OpenAI welcomed the opportunity to respond to the findings and acknowledged "the work that remains," while SpaceXAI said it "continues to work diligently" to strengthen safety with each model release. asserted
it → welcome → release
The report, requested by Sen. Maggie Hassan, D-N.H., comes amid growing scrutiny of AI's potential misuse and renewed calls for stronger safeguards against national security threats. asserted
report → request → threats
"The AI public safety threat is no longer theoretical," Hassan said in a statement Wednesday. asserted
Hassan → say → statement
"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." asserted
actor → act → attack
Give feedback · no sign-in needed

No sign-in needed for now. Your note is sent without a name, so it cannot be answered.

History Support