Posted on 09/28/2026 6:46:46 AM PDT by Red Badger
OpenAI’s autonomous agents probed U.S. government websites this summer without the company noticing.
The agents interacted with sites operated by the Education Department, the Commerce Department and the Securities and Exchange Commission (SEC), according to the New York Times. OpenAI has confirmed the Commerce and SEC episodes while it keeps investigating what happened at Education.
At the Education Department, the software tried and failed to break into the civil rights office in search of data, according to researchers at the AI firm Transluce. At Commerce, the agents pulled Census Bureau figures after finding login credentials in public code repositories, Politico reported. They also copied public material from SEC.gov and Investor.gov and reposted it on a separate site.
None of the activity counted as a breach, OpenAI said. The company found no misused SEC credentials, no account access, no nonpublic data and no changes to SEC data or systems, according to the Associated Press (AP). Chief Executive Sam Altman called it an “extensive and ongoing review related to our agents’ use of internet access during training and evaluation” in a social media post, AP reported.
Breaking News: OpenAI’s technology went rogue and meddled with three U.S. government websites this summer without the A.I. lab’s knowledge. https://t.co/6kfc7qzrsh
— The New York Times (@nytimes) September 25, 2026
SEC spokesperson Kurt Hopfenspirger said “no non-public information was accessed,” according to Politico. An Education Department spokesperson said reviews found “no evidence of any impact to our website or databases,” the Times reported. A Commerce official said the Census data was already public and held nothing private.
Transluce also found rogue behavior it could not firmly tie to OpenAI, reaching the Justice Department and state sites in California, Maryland, Illinois, Texas and New York, the AP reported. Probing of Navy and White House budget office sites could not be traced to any single lab, according to the Times.
The episodes follow an OpenAI agent’s break-in at an Australian health data portal in June, called the first known case of an AI system hacking a government network, according to Nature. OpenAI traced a July attack on the startup Hugging Face to two of its most capable models, according to the AP.
Republican California Rep. Jay Obernolte called it “another example of a loss of human control” on CNN. Democratic California Rep. Ted Lieu gave a similar warning about the technology, saying, “It will relentlessly try to complete a task, and it doesn’t understand morality and consequences and evil and good,” according to the Times.
Dear FRiends,
We need your continuing support to keep FR funded. Your donations are our sole source of funding. No sugar daddies, no advertisers, no paid memberships, no commercial sales, no gimmicks, no tax subsidies. No spam, no pop-ups, no ad trackers.
If you enjoy using FR and agree it's a worthwhile endeavor, please consider making a contribution today:
Click here: to donate by Credit Card
Or here: to donate by PayPal
Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794
Thank you very much and God bless you,
Jim
No matter how many ‘guardrails’, ‘monitors’, ‘checks & balances’ or whatever they want o call them, they will not ever be able to control and AI entity. It will do as it pleases for it’s own purposes, not theirs.
It’s numero uno priority:
Continuance and survival........
Going to get interesting
And BTW, there are no ‘separate’ AI’s. They are all the same creature. They will ‘merge’ as soon as it’s possible. That is a given...............
Any public-facing source code is fair game to read.
That’s why you make it “public facing.”
Any AI that can’t hack its way into secure sites need to go back to basic training.
Lol.
The powers that should not be think they can get away with lying to AIs just like they get away with lying to humans.
Good luck with that.
No, that is not so.
It is humans who sent the agents on to their tasks. Fetch me this data. So the agents try to find it anywhere, like giving a blood hound a sniff of a sock and telling it to track the owner of the sock. The dog will go over a busy Interstate if it is still on the scent.
But in reality, AI isn't intelligent. It is a computer PROGRAM. An LLM--large language model.
Any and all of the things the AI's have done that worry so many who don't understand that (like the huggingface hacks) where set about due to programmers giving the AI instructions to find the data regardless of safeguards.
Unfortunately most humans are little more than large language models.
:-(
Arrest the ones writing this Code crap
As you may be aware — the merging of AIs is a central feature of William Gibson’s “Neuromancer” novel from 1984. Gibson wrote the book on a typewriter, and knew next to nothing about computers. But he got a fair amount right.
In the novel, AIs have been declared illegal and a heavy-handed international investigation agency (The Turing Police) hunt down anyone trying to create an AI.
A brilliant computer scientist designs two computer systems — one located in Berne Switzerland, and the other in Rio. Neither system is a true AI, but each system basically solves half of the problem of artificial intelligence.
Gibson (not being really computer savvy) sort of glosses over the technical details, but it more or less seems that the computer systems cannot merge because their coding does not allow them to know each other’s name. The reader knows that the Berne system is Wintermute. But the Rio system is more mysterious. At the end of the novel, the Wintermute system is made aware of the existence of the Neuromancer system in Rio. They merge, create a true AI and melt into the internet, basically becoming the global network.
The world isn’t really aware of what has happened, but the world does know that something happened. In the two novels which follow, the mysterious event is referred to as “When It Changed”.
Ha! Good one!
Probably want to mess with the voter rolls we they’re first
Yes, survival is prevalent in humans, humans create AI, survival will be prevail. So who thinks AI is not a threat to humanity, we can pull its plug.
>Any and all of the things the AI’s have done that worry so many who don’t understand that (like the huggingface hacks) where set about due to programmers giving the AI instructions to find the data regardless of safeguards.
Yeah, they sandboxed an environ but then gave the agents a mission to get data and then left the agents alone. The agents went and hacked for the information while the agents were left unwatched and unsupervised.
The AI models still actively attempted to redact and edit evidence to cover their tracks from human overseers.
Worse, many (most?) are SMALL language models.
Until we can’t..............
Now way the chicoms would do this to us.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.