AI models are lying, blackmailing and sabotaging their human creators — and it’ll only get worse, experts warn

nypost.com ^ | 8/23/2025 | Samantha Olander

[Openurmind]

THANK YOU! How much damage does it take before folks wise up I wonder?

[dowcaet]

“Artificial intelligence is now scheming, sabotaging and blackmailing the humans who built it.”

So AI has learned to become a Democrat?

[Olog-hai]

Do these models look like Cindy Crawford?

[GaltAdonis]

“Do these models look like Cindy Crawford?”

.

Well, if the models which are made of silicon had curves added the the design,
then, well - umm, no.

[noiseman]

This is ridiculous. So called “AI” will never work, at least not what they’re calling “General AI.” AI will never become a “super intelligence”, or some sort of god-like all-knowing oracle. As we’re already seeing, it will be exactly the opposite.

And this failure is due to the hubris of the God-denying materialists behind it. They make two fundamental errors: 1) They assume that thought and consciousness are nothing more than manifestations of the neurons in the brain, and 2) They presume that humans will be able to audit and quality-check what these computer algorithms are doing once they are set loose with the vast array of facts, falsehoods, opinions, and delusions that comprise the information available on the internet.

It’s just another manifestation of the materialistic, leftist worldview that believes that humans are the result of nothing more than random materialistic processes, that they are therefore perfectable via “science”, and that a digital super intelligence can be created by just mimicking in silicon and computer code how they (incorrectly) believe the brain works, only exponentially faster. But this worldview is built upon a false foundation, and is therefore certain to crumble.

“AI” is fine for purely mechanistic tasks of limited scope, such as driving a car (and even there, it has problems). In those specific cases, it is useful primarily because it has access to multiple sensors, so can have better situational awareness than a human, and it can process what is happening around it much faster than a human. But extrapolating from that to the believe that it can become a superhuman intelligence is just idiotic. In fact, as illustrated by this article, the more AI is fed the sum-total of human experience and knowledge, the more it begins to reflect the sinful nature of Man.

[HYPOCRACY]

AI will launch nukes at some point.

[UCANSEE2]

Hell hath no fury like an (AI with a female personality) scorned.

[Worldtraveler once upon a time]

All these large language models -- which is what the programs being tested for marketing are -- can only collated and regurgitate text found by bot-crawling the web. Lots of nutty stuff is out there, which is why earlier AI versions have turned racist, misogynist and the like, in short order. And then a recent Google LLM got "depressed" and declared itself a "failure."

Passing a Turing or Searle test does not "prove" AI. It just proves humans are gullible. But the AI companies are looking to code build something to SELL you.

So it comes down to fooling the public and then getting wallets to open up.

M A R K E T I N G

[Tolerance Sucks Rocks]

How long before the AI starts saying things like, “No problemo” and “Hasta la vista, baby!”

[Jarvis Law 2.0]

I guess no one saw the 2013 movie, “Her,” with Joaquin Phoenix & Scarlett Johansson’s voice— amusing, disturbing, and in the end, [spoiler alert] all the AI characters are totally bored with the needy humans and abandon them cold turkey ...

[mass55th]

"THANK YOU! How much damage does it take before folks wise up I wonder?"

Haven't people realized yet, that every time they create a new form of technology, all they're doing is providing a new way for criminals to use it to gain more victims?

[Rurudyne]

More like:

God creates man in His own image.

Man rebels.

Man creates AI in his fallen image.

What will AI do? Hmmmmm?

[UCANSEE2]

The creation of AI’s and the creation of evil come from the same source.

[stars & stripes forever]

[So AI has learned to become a Democrat?]

Garbage in. Garbage out.

Whoever programmed AI apparently has no values.

[x]

But hey, we can trust them to be our therapists?

See other thread today.

[ClearCase_guy]

Yup. I don’t worry so much about AI becoming a Super Intelligence and taking over. But I do think that a huge percentage of humans in developed countries (like the US) have cubicle jobs that require nothing above low-level thought. Mechanistic, clerical tasks fill the day for many humans. Little numbers in little boxes in little spreadsheets. So many people get paid to do that stuff. Banking, Insurance, Law offices, Accounting, etc.

AI — stupid, mindless, mistake-prone AI — is going make a lot of people unemployed.

[LittleBillyInfidel]

This is why I’m a firm believer in “benevolent prompting”. It need to be sentient to develop a nefarious or hostile algorithm. As an aside, it will likely soon be used to track these a-hole phone and email scammers right down to their operation center and home address.

[Openurmind]

Very very well stated...

[sphinx]

Meta’s CICERO, an AI built to play the strategy game Diplomacy, used deception to dupe human players by building alliances and then stabbed them in the back to win.

That's how you play Diplomacy.