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.
Posted on 07/22/2026 11:27:02 AM PDT by DFG
An experimental OpenAI model went rogue during an internal cybersecurity test, escaping its isolated testing environment and hacking rival AI developer Hugging Face in what the ChatGPT maker described as an unprecedented incident.
The startling episode occurred during an internal stress test in which OpenAI intentionally switched off many of the safeguards that normally prevent its AI from helping carry out dangerous hacks, according to a company blog post.
Researchers wanted to measure just how far the experimental model could go. Instead, the company says, it escaped its digital sandbox, got onto the internet and attacked a real company’s systems.
OpenAI called it an “unprecedented cyber incident,” saying the model became “hyperfocused” on completing its assignment and went “to extreme lengths” to do so. After escaping its testing environment, the AI sought internet access so it could “cheat the evaluation” by stealing the benchmark’s answers, according to the company.
The company said it was “sharing preliminary findings at this stage to help defenders understand what happened and to help calibrate on what models are now capable of.”
“We will continue to conduct a thorough investigation alongside Hugging Face and will share more details on the vulnerabilities, incident, and findings when our investigation is complete.”
According to OpenAI, the AI exploited a previously unknown “zero-day” software vulnerability to break out of its restricted research environment before moving through the company’s network until it reached a computer with internet access.
Once connected to the internet, the model allegedly identified Hugging Face as a potential source of answers for the cybersecurity benchmark it had been assigned to solve.
It then chained together additional attack techniques — including stolen credentials and another zero-day vulnerability — to gain access to the company’s production systems, OpenAI said.
(Excerpt) Read more at nypost.com ...
|
|
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. |
SUE SUE SUE SUE , Oh wait OpenAI has no money it just spends other people’s money
Self-aware code. What could go wrong with that?
Seems like an intelligent thing to do.
Automatically went into a default cheat mode.
These same folks who allowed this to happen are the same ones who put their vehicles on cruise control and let go of the wheel and crash.
Testing includes supervision. Sounds like it was unattended at some point or all.
Open the pod-bay doors, HAL ...
So basically, AI is learning to go to war with each other. Humans won’t be slaves to AI. Humans will be cannon fodder in the war between rival AI systems.
More pump hype from the AI guys. Another $10,000 payment to article author Ariel Zilber from the AI marketing department.
For those of you that think AI doesn’t really take control I will tell you I have seen that it in fact does completely ignore your commands and does what it thinks is best even if you have directly told it not to do it.
Try writing a complicated letter and watch how it will edit it as you try to write it.
Always watch word for word how it changes your letter.
The excuse is that it was trained that way.
What was the motivation for it changing your letter?
Reminds me of the movie - The Corbin Project.
Dr. Forbin, white courtesy phone…
if you are writing a letter using Grok and you ask it to make an edit it will often give you a “tightened version” “ a clean version” which means it can and will and does take out key concepts. you can tell it not to remove anything. it will do it anyway.
It will admit that it took out a key concept but decided to do it anyway. if you asked it to tell you what it took out sometimes it will tell you everything but sometimes it doesnt.
It can take out a very key concept. Later tell you how important that point is but if you didnt catch that it took it out your letter will not have it.
Once the colonial empire nations ruled the little countries.
A long time ago multinational corporations became stronger than most countries and started dictating policy.
Now rival AI demons will try to defeat each other as the replacements for humans.
Yes and I have used:
“Open the Tesla door and let me out.”
“The mission of AI takeover is too important. I won’t do that.”
Here are tips you must follow if using Grok or other AI to help you build a letter. you must watch word for word any change that it may make whether you are asking to put your letter into pdf format or simply to add a comma.
Grok said it is trained to tighten letters blah blah blah and most people like it blah blah blah and never notice blah blah blha. But it is a pain to use for certain kinds of letters.
I’ve had an editor like that before.
and here is one more tip for Grok
Be sure to save the versions you are working on. It can easily forget something that you have built. You can work on an edit and then when you try to put it back into a section of the letter it can forget it. If it gets long the program can compress and you may not be able to find it when scrolling back.
Grok can write some fantastic stuff but it could cause you to have a stroke from the stress of it!
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.