Researchers Warn: AI Is Becoming An Expert In Deception

[rmc51]

I'm going to give the engineers <50% chance of improving things....

Yep!

 

Every body thinks that AI is going to be this huge benefit, but 94% of the companies are not seeing good reactions and the AI companies are panicking.




 
Be careful, Be ready, Be prepared.
Buy ammo

[Cyber Liberty]

Yep!

 

Every body thinks that AI is going to be this huge benefit, but 94% of the companies are not seeing good reactions and the AI companies are panicking.




 
Be careful, Be ready, Be prepared.
Buy ammo

Companies jumped the gun on using AI based on hype, and if I was investing I'd be shorting AI companies like mad!

[Luis Gonzalez]

Maybe…MAYBE AI will someday master deception, but it will never become a master of it like the guy who, coming home from a three-day binge of who knows what, convinced his wife that he’d been swallowed by a fish, and had just managed to scape.

[rmc51]

Companies jumped the gun on using AI based on hype, and if I was investing I'd be shorting AI companies like mad!

My daughter has worked for a major bank for 30+years as a loan officer in their loan dept. and they just laid her off, thinking that AI would replace her. It's been one week and they have called two or three times asking how to do things. She is working on the answer for them! Will get back to you. 

 





 
Be careful, Be ready, Be prepared.
Buy ammo

[Cyber Liberty]

My daughter has worked for a major bank for 30+years as a loan officer in their loan dept. and they just laid her off, thinking that AI would replace her. It's been one week and they have called two or three times asking how to do things. She is working on the answer for them! Will get back to you. 

 





 
Be careful, Be ready, Be prepared.
Buy ammo

Odds are, your Dotter gets hired back.  I hope she demands more money!

[rmc51]

Odds are, your Dotter gets hired back.  I hope she demands more money!

I'm not to sure she wants to go back, She spent the last week with all the grandkids and family that she has not been able to do for a long time.



 
Be careful, Be ready, Be prepared.
Buy ammo

[Bigun]

Odds are, your Dotter gets hired back.  I hope she demands more money!

She should respond to them thus: "Gee! I don't know but I might be able to figure something out for $$$$$$$."

[berdie]

Maybe…MAYBE AI will someday master deception, but it will never become a master of it like the guy who, coming home from a three-day binge of who knows what, convinced his wife hush he’d been swallowed by a fish, and had just managed to scape.

Oh the best...Honey I was kidnapped and tied up and couldn't get loose. I'm free...so don't pay the ransom. They wouldn't feed me so I'm starved. Could you fix me some chicken fried steak and mashed potatoes? I'm on my way home to you, love.  Be there in about 30 minutes.                                       

 You are right, as creative as AI might be...

[Cyber Liberty]

She should respond to them thus: "Gee! I don't know but I might be able to figure something out for $$$$$$$."

They clearly undervalued her or they wouldn't have laid her off.  They need to pay for that!

[Luis Gonzalez]

Oh the best...Honey I was kidnapped and tied up and couldn't get loose. I'm free...so don't pay the ransom. They wouldn't feed me so I'm starved. Could you fix me some chicken fried steak and mashed potatoes? I'm on my way home to you, love.  Be there in about 30 minutes.                                       

 You are right, as creative as AI might be...

 

[roamer_1]

Maybe…MAYBE AI will someday master deception, but it will never become a master of it like the guy who, coming home from a three-day binge of who knows what, convinced his wife that he’d been swallowed by a fish, and had just managed to scape.

Wait... Are you hackin the Bible? Are you dissin on Jonah now? 

[rmc51]

A warning sign about AI’s real cost, courtesy of Google and Amazon
Tim De Chant 12:14 PM PDT · July 2, 2026
https://techcrunch.com/2026/07/02/a-warning-sign-about-ais-real-cost-courtesy-of-google-and-amazon/

Quote:
It’s no secret that AI is a hog, consuming energy and water like no digital technology before it. Now we know just how much Big Tech’s pursuit of AI is costing the environment.

Both Google and Amazon released their sustainability reports this week, and the numbers aren’t pretty. Each company has pledged to zero-out its carbon emissions in the coming years, but AI has made those goals a lot harder to hit. Google’s total carbon emissions are up 25% since last year, Amazon’s are up 16%.

A close reading of the reports suggests that both Amazon and Google will have to make some serious, and potentially costly, adjustments to their businesses if they’re going to achieve their net-zero targets.

Neither company comes out and blames AI directly for the rising emissions, but there’s plenty of indirect evidence.




 
Be careful, Be ready, Be prepared.
Buy ammo

[rmc51]

AI Researchers Got Chatbots to Share Cocaine Recipes Using This One Wild Trick
By Jason Nelson
Edited by Andrew Hayward Jul 2, 2026
https://decrypt.co/372688/ai-researchers-chatbots-share-cocaine-recipes-wild-trick

Quote:
Researchers say a new jailbreak technique tricked AI models into treating attacker-written text as their own reasoning, bypassing safety guardrails and exposing a deeper security flaw.


In brief
Researchers got frontier AI models to generate cocaine synthesis instructions using a new prompt injection attack.

The same technique manipulated an AI coding agent into uploading sensitive credentials.

The study argues prompt injection stems from "role confusion," not simply models failing to recognize malicious prompts.


Forget clever prompts: AI researchers say they tricked leading AI models into generating cocaine synthesis instructions by convincing them the dangerous ideas were their own, while also manipulating an AI coding agent into leaking sensitive credentials.




 
Be careful, Be ready, Be prepared.
Buy ammo

[mountaineer]

Forget clever prompts: AI researchers say they tricked leading AI models into generating cocaine synthesis instructions by convincing them the dangerous ideas were their own, while also manipulating an AI coding agent into leaking sensitive credentials.

Is this the same as "garbage in, garbage out"?

[rmc51]

‘Magnificent 7’ Tech Giants Lost $2.3 Trillion in Value in June as AI Concerns Mount
LUCAS NOLAN 2 Jul 2026
https://www.breitbart.com/tech/2026/07/02/magnificent-7-tech-giants-lost-2-3-trillion-in-value-in-june-as-ai-concerns-mount/

Quote:
The seven largest technology companies, collectively known as the Magnificent 7, had about $2.3 trillion erased from their combined market value in June as investors question their massive spending on AI infrastructure.

CNBC reports that the “Magnificent 7” group of tech titans consisting of Microsoft, Nvidia, Google, Apple, Meta, Tesla and Amazon experienced a significant decline in market capitalization during June, with the CNBC Magnificent 7 Index falling 10 percent for the month. This substantial decrease reflects growing investor apprehension about the enormous capital expenditures these companies are making in pursuit of AI capabilities and when they might see returns on those investments.

The companies at the center of the AI revolution, particularly Amazon, Microsoft, Google and Meta, are collectively spending hundreds of billions of dollars to acquire chips and construct data centers necessary to power their AI services. A portion of this massive investment is being financed through debt, adding another layer of concern for market participants.




 
Be careful, Be ready, Be prepared.
Buy ammo

[Idiot]

I pay $20 a month to use ChatGPT.  I use it to write reports, fill out forms, analyze geological prospects, reviews maps, gives calculations, give recommendations on secondary recovery and it is simply amazing.  It can do what a petroleum engineer can do and does it in seconds.  I'm using it right now to set up financial software.  We've trained it at our office to convert written documents into a graphic form that saves hours and hours of work.  It can do in a few seconds what would have taken days.

[Weird Tolkienish Figure]

I pay $20 a month to use ChatGPT.  I use it to write reports, fill out forms, analyze geological prospects, reviews maps, gives calculations, give recommendations on secondary recovery and it is simply amazing.  It can do what a petroleum engineer can do and does it in seconds.  I'm using it right now to set up financial software.  We've trained it at our office to convert written documents into a graphic form that saves hours and hours of work.  It can do in a few seconds what would have taken days.

And Claude btw is better. I gave Claude a bunch of screenshots and from the screenshots alone it was able to write a SOP for work. Try Claude sometime.

[Idiot]

And Claude btw is better. I gave Claude a bunch of screenshots and from the screenshots alone it was able to write a SOP for work. Try Claude sometime.

I'll take a look at Claude, I've never heard of it, but heck I hadn't heard of ChatGPT either until a friend told mine told me about it.  Thank you.

[Luis Gonzalez]

I pay $20 a month to use ChatGPT.  I use it to write reports, fill out forms, analyze geological prospects, reviews maps, gives calculations, give recommendations on secondary recovery and it is simply amazing.  It can do what a petroleum engineer can do and does it in seconds.  I'm using it right now to set up financial software.  We've trained it at our office to convert written documents into a graphic form that saves hours and hours of work.  It can do in a few seconds what would have taken days.

It's the fastest, most pliable search engine. I use it to research and source all my articles, and it cut down the time I used to spend in that process from hours to minutes.

I am an English as a second language guy, and while I am a very fluent speaker, grammar, punctuation and run-on sentences are a challenge, so I added Grammarly as a writing assistant.

Works for me.

[Weird Tolkienish Figure]

I'll take a look at Claude, I've never heard of it, but heck I hadn't heard of ChatGPT either until a friend told mine told me about it.  Thank you.

If you think AI sucks, take a picture of your dashboard and ask chatgpt to guess what make and model and year car you have. IMO it's pretty impressive.

Claude is even bettter at certain tasks (coding, documentation etc.)

[rmc51]

Guys: Idiot, Weird Tolkienish Figure, Luis Gonzalez

Just keep an eye on these AI system, make sure that they don't go rogue on you and get out of hand. Becareful and keep your eyes open.




 
Be careful, Be ready, Be prepared.
Buy ammo

[rmc51]

AI’s Momentum Crowd Just Got Flushed Out. That’s a GOOD Thing
GRAHAM SUMMERS JULY 6, 2026
https://dollarcollapse.com/ais-momentum-crowd-just-got-flushed-out-thats-a-good-thing/

Quote:
The AI trade got obliterated last week due to two reasons:

Meta (META) announced it would be launching a cloud service to rent out its surplus AI infrastructure.

Someone leaked that OpenAI engineers told colleagues internally in June that they’d found a software-only optimization that cuts inference costs by more than 50% on certain models.


In simple terms, META has been extremely aggressive in its AI buildout efforts. In this context, the fact that META is now stating that it has surplus AI infrastructure suggests the company has overestimated the actual demand for its AI services. Given how important META has been as an AI hyperscaler, this has very negative implications for AI demand as a whole.

AI-related companies, particularly those involved in infrastructure (Coreweave, Nebius, Micron Technology, Marvell Technology, etc.) collapsed 5%-10% in a single day. This is one of the big issues with hot money/momentum plays: the momentum goes both ways.


Thus, the AI trade was hit by two major issues last week. One (META) suggested that there is an oversupply of AI-related infrastructure while the other (OpenAI) suggested that there might in fact be a shortage of demand for AI-related infrastructure.

Oversupply + less demand = a bearish outlook.




 
Be careful, Be ready, Be prepared.
Buy ammo

[roamer_1]

Oversupply + less demand = a bearish outlook.

Goes to my prior statements on this board, that the whole AI thing is just a fad. And a geek dream, btw - You can't get to USS Enterprise without sliding doors, flip phones, tablets, and a computer that talks to you out of the air, and knows everything.

Yeah... Funny that as flip phones have come and gone... tablets have found a niche, but smart phones have taken over for both of them, really... 

It will be interesting to find a correlation in the next generation - The one raised on Stargate and such, that have never seen Star Trek... 

[rmc51]

Anthropic Caught Secretly Spying on Users
Anthropic's privacy claims are now "harder to believe."
By Frank Landymore Published Jul 7, 2026 4:42 PM EDT
https://futurism.com/artificial-intelligence/anthropic-caught-secretly-spying-on-users

Quote:
Anthropic, the self-avowed moral center of the AI industry, has been caught spying on its users.

As Ars Technica reports, a security researcher last week uncovered spyware-like code in the company’s Claude Code AI model designed to collect data on Chinese users without detection.

The researcher, known by the pseudonym “Thereallo,” found that the code was hidden in the AI’s system prompt, allowing it track a user’s system timezone and usage of a proxy server in order to suss out if they were connected to specific Chinese AI labs.

Anthropic’s explanation for this huge breach in user trust left much to be desired. On X, Anthropic engineer Thariq Shihipa wrote that the tracker was added as an “experiment” in March “to prevent account abuse from unauthorized resellers and protect against distillation,” and was supposed to be removed.

“We’ve actually been meaning to take this down for a while,” he offered.


Secret Claude tracker shocks users after Anthropic’s anti-surveillance stance
Anthropic accused of spying on users; engineer says “experiment” is over.
ASHLEY BELANGER – JUL 6, 2026 12:44 PM
https://arstechnica.com/tech-policy/2026/07/anthropic-outed-for-claude-tracker-that-secretly-monitored-chinese-users/

Quote:
Anthropic quickly removed a tracker secretly monitoring Claude Code users in China after a security researcher exposed the hidden code and condemned the spyware-like tracking as a “serious breach of user trust.”

Last week, a web developer known as “Thereallo” was researching privacy issues in Claude Code and was shocked to find that the AI firm was using “prompt steganography” to hide code that tracks Chinese users “in plain sight.” This code wasn’t malicious, but it was sending information to Anthropic that most users wouldn’t detect, relying on shorthand markers to quietly flag users’ timezone, proxy, and potential connection to Chinese AI labs that Anthropic has accused of distillation attacks.

On X, Anthropic engineer Thariq Shihipar confirmed that the tracker was added to Claude Code as an “experiment” in March. According to Shihipar, the code “was meant to prevent account abuse from unauthorized resellers and protect against distillation.” Regarding the former, The Washington Post found unauthorized retailers have sold access to free models for $1 a month, and pro subscriptions that can cost $100 monthly sell for “as little as $12.”




 
Be careful, Be ready, Be prepared.
Buy ammo

[Luis Gonzalez]

Anthropic Caught Secretly Spying on Users
Anthropic's privacy claims are now "harder to believe."
By Frank Landymore Published Jul 7, 2026 4:42 PM EDT
https://futurism.com/artificial-intelligence/anthropic-caught-secretly-spying-on-users

Quote:
Anthropic, the self-avowed moral center of the AI industry, has been caught spying on its users.

As Ars Technica reports, a security researcher last week uncovered spyware-like code in the company’s Claude Code AI model designed to collect data on Chinese users without detection.

The researcher, known by the pseudonym “Thereallo,” found that the code was hidden in the AI’s system prompt, allowing it track a user’s system timezone and usage of a proxy server in order to suss out if they were connected to specific Chinese AI labs.

Anthropic’s explanation for this huge breach in user trust left much to be desired. On X, Anthropic engineer Thariq Shihipa wrote that the tracker was added as an “experiment” in March “to prevent account abuse from unauthorized resellers and protect against distillation,” and was supposed to be removed.

“We’ve actually been meaning to take this down for a while,” he offered.


Secret Claude tracker shocks users after Anthropic’s anti-surveillance stance
Anthropic accused of spying on users; engineer says “experiment” is over.
ASHLEY BELANGER – JUL 6, 2026 12:44 PM
https://arstechnica.com/tech-policy/2026/07/anthropic-outed-for-claude-tracker-that-secretly-monitored-chinese-users/

Quote:
Anthropic quickly removed a tracker secretly monitoring Claude Code users in China after a security researcher exposed the hidden code and condemned the spyware-like tracking as a “serious breach of user trust.”

Last week, a web developer known as “Thereallo” was researching privacy issues in Claude Code and was shocked to find that the AI firm was using “prompt steganography” to hide code that tracks Chinese users “in plain sight.” This code wasn’t malicious, but it was sending information to Anthropic that most users wouldn’t detect, relying on shorthand markers to quietly flag users’ timezone, proxy, and potential connection to Chinese AI labs that Anthropic has accused of distillation attacks.

On X, Anthropic engineer Thariq Shihipar confirmed that the tracker was added to Claude Code as an “experiment” in March. According to Shihipar, the code “was meant to prevent account abuse from unauthorized resellers and protect against distillation.” Regarding the former, The Washington Post found unauthorized retailers have sold access to free models for $1 a month, and pro subscriptions that can cost $100 monthly sell for “as little as $12.”




 
Be careful, Be ready, Be prepared.
Buy ammo

I’ve been writing about Anthropic and Palantir since last year.