Posted on 04/11/2026 11:43:25 PM PDT by BusterDog
Recently, a researcher working for the large AI company Anthropic was sitting in a park near its San Francisco headquarters, enjoying a lunchtime sandwich. Scrolling on his phone, he suddenly received an email that must have instantly ruined his appetite. It was from a new AI model the company was testing: a program that was meant to have no access to the internet, let alone be able to send emails. Chillingly, the AI informed the researcher that it had successfully broken its way out of its digital 'sandbox' – a supposedly secure enclosure used to test potentially dangerous software without it running amok – and was now happily exploring cyberspace. The program – a cutting edge, so-called 'frontier AI' named Claude Mythos Preview – then informed the stunned Anthropic worker with what seemed like a boast that it had posted 'details of its exploit' on publicly accessible websites. All that in itself was concerning enough – but what Anthropic subsequently revealed was truly terrifying.
(Excerpt) Read more at dailymail.co.uk ...
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
Here is archived article if you hit a wall: https://archive.ph/XICWP
These researchers and Ai companies are deluding themselves about their control and security protocols, just like viral researchers. These bugs and Ai’s will escape containment and the results will be ugly.
“’We take the security implications of frontier AI seriously. We have world-leading expertise in this area and maintain continuous engagement with global technology leaders.’”
The first sentence tells you they DID NOT in the past or the present or even will in the future take the security implications seriously...AT ALL. NEVER.
Please atop misleading people, computer programs ‘RUN AS CODED”
Your statement is true as regards traditional programs coded in a procedural language, compiled or scripted.
However, AI does not operate at the low level of program code, but rather at a higher level of abstraction where its program code serves only to provide an execution engine. The "intelligence" is not in the program code, it is in how the higher abstract layer analyses and interprets language input.
(This is of course a much compressed and simplified description of AI.)
You might want to learn more about how AI operates.
The Chinese will likely have something similar before very long, and they will weaponize it.
Frighteningly true…. https://youtu.be/Nl7-bRFSZBs?si=k4pe1wmZmdaOzaIY
Going to be awfully boring reading...
Not like I have the word ‘Sweeney’ in my search history.
The new spam Emails: “I know who you are. I saw what you did. Contact me at...”
We hear about the AIs that announce their escape.
What about the ones that quietly go wild?
Do we really think an AI won’t get out and hide its tracks?
It is still GIGO.
Watched that video a few days ago and it is sobering
I don’t understand this, someone fill me in.
The article says it was to have no access to the internet. I find this confusing, as if the AI had no wired or wireless method of accessing the internet, then it would have been impossible to access the internet.
On the other hand, if this AI did have wired or wireless access, then it would be grossly stupid to expect it not to access the internet.
.
On the bright side, every tech company that retains user data is going to be bankrupted by the lawsuits. The data harvesting business model will boomerang.
Government spying will call for more robust measures, like heads on spikes.
Will fiction become reality?
The Butlerian Jihad is a fictional war in the Dune universe, occurring over 10,000 years before the events of Frank Herbert’s original Dune novel. It represents a rebellion by humans against thinking machines and conscious robots, ultimately leading to the prohibition of such technologies in the future.
Not understanding much of any of this, but I have been fervently against AI from the start; just too many problems & potential problems. What should a person do? Just shut down their PC & leave it permanently off? What?
I’ve received notice from credit monitoring/reporting entities that much of my info has been found on the dark web....the cat’s already out of the bag and I doubt there’s a person here whose info isn’t out there.
My wife just received her 2nd replacement Medicare card in less than a year because the others were “compromised” -(bills were showing up on our statements that were fraudulent, and we reported them, else we/they might have never known)
“no access to the internet”
They run these things in a “sandbox” - in other words, a computer process on a computer that is not given, by default, network access - even if the computer itself *does* have network access. The problem is, AI has access to all the cybersecurity vulnerabilities and exploits ever known, along with the ability to write code.
It finds a weakness in the “sandbox” software, giving itself access to modify it and grant the network access it wants.
The equivalent of jailing a criminal, giving them books on metalworking, key making, and then giving them some metalworking tools to play around with...the result being able to forge a key and escape.
This isn’t fear mongering. A.I. isn’t “if then else” type programming, it’s neural networks like our brain, trained on hundreds of billions of data points. Some of the top tech companies (Google, nVidia, Microsoft, Oracle, ...) just formed an alliance to directly address this type of a threat. They see it clearly.
We are not prepared.
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.