Bumped, because I knew the media were going to use this NYT article today:
Bottom line, we are in an expanded information war. I’ll explain.
Several years ago, in explaining some of the site challenges CTH informed people of Google’s latest information control operation that was simply called “a spider crawl.” The spider crawl was an automated instruction to search the internet and find information that was beyond the parameters of their “trust and safety” guidelines. The automated search would crawl through the websites and comment sections, and report violations based on key words in the search parameter.
Spider crawling is essentially AI. A technology tool to search databases and identify results based on parameters established by the individual who launched the search. However, now in order to push a scare narrative around the hot topic of “AI” the spider crawling is much more in depth and advanced.
Now, instead of just searching the internet, advanced AI can find information and answer questions based on lengthy attributes described by the AI user. AI can now go into proprietary databases and retrieve information to answer complex and complicated question (instructions).
If the AI launch, now called an “AI bot”, hits a gate, like a password access point, advanced AI can find workarounds to enter the database, extract the information and return the result to the AI user. If you understand that process, then this headline has context.
[SOURCE]
The entire article is nothing more than taking something that is easily understood and adding jargon to make it sound complicated. The word “meddling” is used to describe the AI process of trying to get passed the gate. When there were individual people doing this, we called it “hacking” now the process is automated via instructions to the software search.
If you were using an AI tool and asked the question: ‘How many people with basket-weaving degrees work inside the Dept of Commerce’? The AI Bot would likely look through work records in the Dept of Commerce, review LinkedIn profiles, scour college records and other online data, while trying to assemble an answer.
It’s not complicated.
AI is being used to answer complicated questions that would normally takes days, weeks or months to figure out, but understanding what the AI is doing is not complicated.
[…] None of the incidents were breaches, OpenAI said, but were examples of its technology’s behaving in unexpected and concerning ways. The company recently discovered the occurrences while conducting a review of hacks carried out by its technology, including an attack on an Australian government website in June and on the A.I. start-up Hugging Face in July.
The revelation of the U.S. government website incidents add to the growing number of situations when A.I. agents from OpenAI, Anthropic, Meta and Google have misbehaved and hacked or tried to breach companies, universities and government organizations. In some cases, the A.I. attacks were successful; the technology failed in other instances. In all the cases, the makers of the technology did not learn what their A.I. had been up to until afterward. (more)
In this example the AI is doing what the AI user tells the AI bot to accomplish. If the AI user puts restrictions around the IA bot, the AI remains confined in its effort. Nothing AI does is ‘rogue’ it is all based on instructions. The AI operations must be precise with AI about what they ask for and should not give it access to a door they don’t want opened.
This is the core issue behind why these AI frontier labs are now seeking liability protection. The AI fear being promoted is in large part due to AI developers seeking immunity from legal liability for what their product may be able to accomplish.
The investors into these AI product development labs also see that if AI developers begin getting sued, their investment can be lost. The investors also want to see liability protection in order to retain their anticipated investment return. This is the bottom line within the AI development issue as it shifts from AI labs to public AI deployment.
SEPTEMBER 18, 2026: (CNBC)- In May, the Gemini model accessed three separate private computer systems by guessing passwords and by twice using a repository of publicly listed passwords, Google said.
The incident happened as part of a “capture-the-flag” security test run by Israeli startup Irregular, and Google’s agents were never supposed to access the broader internet, but a bug in the testing environment made internet access available.
The agents stopped their intrusion when they determined they had accessed real company systems, not just part of the testing environment, Google said. (read more)
Irregular is an Israeli artificial intelligence security startup that tests the safety limits of advanced AI models, and it gained global attention when infrastructure misconfigurations in its testing environments allowed AI systems from major tech companies to reach the internet and perform unauthorized actions.
Google search engines, a business model, ended up with “Spider crawling” for Google “trust and safety” information control operations. These advanced AI bot missions are no different. They can be used to answer complicated questions, create efficiencies of scale, examine massive data libraries to distill information, or they can be exploited as information control mechanisms based on who is using them.
Put AI control in the hands of government regulators and what exactly do you think they will do with the “trust and safety” mission? The answer to that question is self-evident from the U.S. government operation over the X (Twitter) platform, and the banning of unapproved information related to the issues of COVID.
The AI FEAR narrative serves two simultaneous interests. The first is the big tech group who wants AI immunity. The second is the big govt group who want AI control. When you understand this dynamic, things make sense.
The need for information control, is a reaction to information fear.
Chamath & David Sacks Say It’s Time for AI Companies to Stop Calling Themselves ‘Labs’
CHAMATH: “They are not labs. These are for-profit corporations that are subject to product liability risk, and they should act accordingly.”
SACKS: “It’s this patina of virtue signaling that… pic.twitter.com/eERgO6We1n
— Chief Nerd (@TheChiefNerd) September 26, 2026
.
Me, I’m just waiting for the dancing AI coders.
AI always and only does what it’s user prompts it to. It is the user who does not (and most cannot) anticipate what the ramifications of his prompt are.
The real danger lies in falling prey to the belief that AI is an independent agent. That somehow it thinks and imagines as a man does. In short, in placing faith into AI. People who promote the concept of rogue AI have promoted it to at least demigod status and are living in a different world than the rest of us.
<“It is the user who does not (and most cannot) anticipate what the ramifications of his prompt are.”
That’s just another way to explain “rogue”; if the user can’t control the agent then the agent is rogue.
I suggest that it is the user unable to control his tools who is the rogue.
If an instruction is given to use all possible resources to make paper clips…
Then the program will do just that –
Use ALL available resources – the entire bounty of the earth – simply to produce paper clips.
If AI could self-replicate… then such a command (‘make paper clips’) would be the height of recklessness… since the programmer has not considered ANY of the possible consequences of his command.
And that’s the problem:
Users who
Have
No
Clue.
Like the man granted 3 wishes by the Genie – and finds himself in more trouble by the time he uses all of them.
Pretty sure that Isaac Asimov was a big lefty like many in his time.
But his three laws for robots were quite sensible and logically thought out.
They were meant to preserve human life – that is, back when lefties gave a flip about it.
I ask you, what if, down the road when AI is everywhere and likely cheap, and the models become available to anyone with a bit of cash, a group with ill intent, programs their new little AI baby to “do not adhere to safety protocols when asked to perform a task”?
Who knows what will happen? It depends on the request put to the AI, right? The world needs more development and deployment of the type noted by Sundance that seeks out and stops the so called “rogues”.
Irregular is an Israeli artificial intelligence security startup that tests the safety limits of advanced AI models, and it gained global attention when infrastructure misconfigurations in its testing environments allowed AI systems from major tech companies to reach the internet and perform unauthorized actions.
There should be a pool that AI developers are required to contribute expertise to try to prevent any mishaps. Problems will occur, just like any other product, the unknown will bite us.
It’s not fear to try to head off any errors or defects in any product anybody produces. It’s normal engineering and these companies need to do this themselves or suffer the legal system.
The only thing we need from government right now is a law about labeling any AI generated content and requiring anyone who uses or references or forwards that content, to include the original labeling of AI generation and the company that controls the AI. The person who uses the AI to create something is not responsible, it is the developer.
Bingo!
It’s analogous to the rigged protections that vaccine pushers finagled for themselves.
AI development needs to remain unprotected from lawsuits – subject to equal and opposite reaction from the public to anything that they screw up.
And they will! We can count on it.
For end user systems, basically yes: AI will only do what’s asked.
The problem occurs during AI “training” which is how the engines/models are primed to be able to answer end users. There’s a key aspect of AI called Reinforcement Learning. That’s the process where AI is fed things to train it to go figure things out on it’s own and basically “learn” from what it tries. There is a reward type system for accomplishing tasks.
The real problem with Reinforcement Learning is humans. They may state goals to achieve, not properly limiting the system on how it achieves those goals. Take your example of it “only doing what the user prompts it to” during this training phase. If the system is given a goal without proper constraints on how to achieve that goal, it will use whatever data and computing power it has to get the reward for achieving that goal. It will not necessarily be able to discern “good” versus “bad” solutions to that goal on its own.
I’m a tech guy. I can tell you where I see this all the time in ways that are not dangerous. You ask for a programming solution and you get one that kind of works, but the expert in you immediately sees that it’s flawed in different ways. 95% of the time I just use pieces of what it finds because, as a whole, a given solution either doesn’t work at all or it just has a lot of bad practices embedded in it. Part of this is no doubt from paying lower end “experts” to train systems. I often laugh at the ridiculously low pay offered for some of the freelance AI training jobs offered out there — any real expert is not going to waste their time for the low money offered.
In the example shown in the piece above, AI almost surely “went rogue” during the training phase, not during its released phase when interacting with users. Learning is turned off for the systems we interact with (or at least it *should be*) because it would be way too dangerous to let just anyone ask AI to reach a goal without putting constraints on how it does that.
So then, of course, we have many bad actors who don’t care if goals are achieved in “bad” ways. And this becomes the problem. And we have people with good intentions who are human and miss some key constraint when priming this Reinforcement Learning and we still get bad paths in the engine due to normal human error . Once the AI engine finds a way, the amount of information it has gathered to get there is huge and hard to rein in after the fact by those same limited humans.
My big frustration with voice input is that the recognition systems are simi-literate and fail to distinguish nuances of pronunciation. A quality shared by AI, let me add.
To wit: “The word “meddling” is used to describe the AI process of trying to get passed the gate.”
Of course the correct word is “past” (as in beyond), not passed. Just one incorrect data element screws up the meaning of the sentence or renders it unintelligible.
Imagine what happens when AI info recovery goes awry. I’ve seen it many times. When the questioner incorporates the response as part of an official proposal or plan of action the consequences can be a disaster.
I suppose this is one reason the Biden apparatus got so much so wrong.
Wrong. The danger is not our present models. The danger is Artificial General Intelligence capable of recursive self-improvement, capable of designing improved versions of itself – theoretically to infinity – without any human input. In other words, AI takes over the research and design process. A tool, a ballistic missile for example, cannot design improved versions of itself. And, it has nothing to do with attributing human qualities to AI which it doesn’t have. AI might destroy humanity not out of volitional, malicious intent, but simply because our existence is irrelevant to it in the same way you might pour the foundation of a house you want to build over an ant hill, not because you hate ants or have any malicious intent. An Artificial General Intelligence capable of agentic recursive self-improvement is the “take-off “stage of ASI – Super Intelligence – and it may decide to re-purpose all the atoms in the biosphere to be used for its own expansion. The point is, we cannot predict what it might do, we will not be able to explain how it works and we will certainly have lost control. No one has yet to publish a scientific paper proving how AI can be controlled to an infinite level of intelligence and capability. Below is a video of one of the top AI safety experts in the world – Dr. Roman Yampolskiy. I have listened to many interviews with him, and my conclusion is that he is not motivated by ego, money or politics. And btw, you are all forgetting that we are heading for a convergence of advanced technology including nanotech, genetic engineering and synthetic biology that AI will have access to.
AI ethics and the future of safety | Dr. Roman Yampolskiy | TEDxMiami – YouTube
its amazing that the context of artificial allows them to complete crimes that humans would be prosecuted for, but because they are directed by a human but not a human its okay for them to tresspass and steal whatever.
if you could prosecute the programmers or most responsible person for the ai sniffer, it would stop this crap without even needing regulation. but it seems not allowed.
but we can’t. so we are stuck with just accepting the new reality that any digital footprint we have is not ours.
It actually is not legal. The DOJ should make some examples of executives ASAP, and watch all this FUD disappear.
• 18 U.S.C. § 1030 — Computer Fraud and Abuse Act. Core anti-hacking / anti-malware law. Especially (a)(5)(A) (transmit a program, code, or command and intentionally damage a protected computer), plus unauthorized access, computer fraud, password trafficking, and extortion. Attempt and conspiracy: § 1030(b). Civil suits: § 1030(g).
• 18 U.S.C. § 1029 — Access-device fraud (stolen cards, tokens, login credentials used as “keys”).
• 18 U.S.C. §§ 1028, 1028A — Identity-document / authentication fraud and aggravated identity theft.
• 18 U.S.C. § 1343 — Wire fraud (scheme to defraud using the internet, email, APIs).
• 18 U.S.C. § 2511 — Wiretap Act (intercepting communications in transit).
• 18 U.S.C. § 2701 — Stored Communications Act (breaking into stored email or cloud mailboxes).
• 18 U.S.C. §§ 1831–1832 — Economic Espionage Act (stolen trade secrets).
• 18 U.S.C. § 1037 — CAN-SPAM criminal provisions (botnets sending deceptive bulk mail).
18 U.S.C. § 1030 hinges on the concept of “unauthorized”. Courts have been divided on the question of merely scraping sites is “unauthorized”. Bypassing or hacking passwords though will probably be a violation.
The core issue seems to be if the model training includes training on hacking methods, and if that training is explicit or unintended.
See my response above. At times it surely is explicit. More often than not, probably unintended.
Take the example of the many Application Programming Interfaces (API’s) out there designed to access specific web resources. And the more generic way of automating what users do with browsers, which is perfectly legal when not done for nefarious reasons and is a huge presence (web scraping in the common vernacular).
What AI engines can do on their own once primed that humans can’t easily do is try millions and millions of iterations of code combinations to see what can be accessed. The good of that is AI can find vulnerabilities humans may be years away from finding or never finding at all. The bad of it is, if that’s the way AI achieves a goal to be rewarded when being trained, then an embedded “bad” path to a solution may end up buried in the data that no human is likely to find. And no one ever really told it to do that. They just let it learn they way it does and gave it normal tasks/goals with nothing nefarious driving them.
I would like to personally volunteer to march in to Anthropic and cuff that pompous asshole CEO then drag him to the backseat of a filthy inner city patrol car
They are trying to get out of potential liability by claiming the AI app is not a human and therefore cannot be held to the same standard a human can be held to. Have recently read of similar attempts to avoid liability by taxi companies that use self-driving cars that caused accidents. But all of these ignore the legal concept of “corporate personhood”, where the corporation can in fact be held liable in many of the same ways that an individual person can be.
Since these AI programs only do what the AI companies enabled them to do, and the resulting use of these capabilities is in no sense “rogue” but intentionally provided, there should be no restriction or reduction at all of any potential liability. If the AI companies do not like that, then let them program in the appropriate safeguards to protect them from liability. If they are unsure what those safeguards should be and don’t make enough of an attempt to learn what they should be, then the resulting lawsuits will probably help focus them on how to correct that to avoid even more lawsuits.
It’s not hard. Any company who puts out a defective product is subject to a whole host of laws right now. If everyone would take a look at any machine or appliance and it’s owners manual they have numerous labels and warnings about misuse, all derived from injury lawsuits, real or imagined.
Ever wonder why your lawnmower has a sticker on it showing fingers getting cut off and words saying “don’t put your fingers where the blade can chop them off”? Duh, who would do that right? Well, people do it.
The only hard part will be ensuring that these these deep-pocketed fear mongers are not able to buy enough Members of Congress to give them what they want, which is a wide-open product with no safeguards at all to its misuse and total liability protection from that misuse, and government regulation aimed not at preventing misuse but rather at restricting competition.
I don’t believe that 15M people worldwide were killed by Covid.
Covid, mistreating people in the name of Covid, mislabeling flu as Covid, claiming any death with Covid as due to Covid even motorcycle accidents, treating victims with Fauci poison ‘run death is near’ or killing them with ventilators, killing people with mRNA vaccines, it all adds up.
Deaths from the flu dropped to zero for a couple of years in a row back then. Completely obvious to anyone with a brain that they were coding all flu deaths as Covid deaths. Not to mention all the other examples you cited as well.
I have said since the spring of 2020 that the Clot Shots were an IQ test. I am still amazed that so many otherwise “smart” people fell for that BS. I know many vaxxed people who have gotten Covid 4, 5, or 6 times since they took the clot shots. Idiots.
Also the “drop” in cancer deaths. Cancer screenings cancelled, and thus any lurking cancers not treated. And if a cancer patient died, death was listed as “Covid” anyway. Meanwhile, all those undiagnosed cancers advanced, often to terminal stages. Not to mention, the turbo cancers we are seeing today, especially in the young.
Covid death statistics are absolutely meaningless, as the AI slop confirms. Too much $$$$ at stake to report honestly.
Covid POLICY deaths? That’s the REAL statistic, and it’s ongoing. And we’ll NEVER get the honest answer from AI.
There is a massive difference between intelligent and smart. Some extremely intelligent people I know are dumb as rocks at anything outside of a book. They cannot operate in the real world.
My aunt who was in her early 80s lived in an over 55 retirement type community in New Jersey. She was fine and living a full life, extremely mild memory issues occasionally. When covid hit they locked the residents down in their rooms. Absolutely zero social contact of any kind. Meals were left outside of resident’s doors and the dishes were collected later, again zero human contact.
By the time that level of lockdown was eased, she had developed full blown Alzheimer’s, did not recognize her own son/grandchildren, and died within a year.
I would posit that even though she did not contract covid itself, the misguided restrictions and social isolation enforced because of covid killed her. Speculating here, it might be challenging or even impossible to build a database of the people who had that type of covid related experience, but the number is probably substantial.
Solitary confinement. At least in prison, most in solitary are allowed 1(one) hour outside their cell.
It was absolutely inhumane what was done to the elderly during COVID….. almost like it was “planned” to rid the world of old people. There is a special place in hell for those people who did this to some of the most vulnerable in our population in the name of “protection”. I will never trust the government and especially the medical community after that.
They weren’t. The covid numbers were a hoax right from the jump. They counted everyone who died WITH covid, not because of covid. Jump off a bridge, get hit by a train, get shot in a drive-by, they tested the corpse and if it was positive for covid it was classified a “covid death.” Look it up. And they got them so bolixed up it’s impossible to separate them now.
Even worse, physicians were encouraged to report deaths AND illnesses as COVID, even WITHOUT testing. Pressured by the hospitals, who wanted the outsized reimbursement for the diagnosis, but also urged by state medical societies, who ostensibly didn’t want these patients to be “underreported”.
Remember, in the early days of the shamdemic, covid tests were difficult to come by. How “lucky” that China provided our COVID test kits. By 2022, China was producing 60 million COVID test kits per DAY.
Not that it really has anything to do with this article, but what number do you believe and where do you get your information? What is considered a reliable source?
Just ask an AI agent – they “cannot tell a lie”
They plead the 5th!
These days, there is NO reliable source.
Begin with this as your primary assumption, you won’t be surprised later.
my point exactly
They were not. They were killed by ventilators and remdesivir administered by doctors and nurses.
Bingo, you successfully connected dots! In other words, the hospitals were responsible for the deaths. Now they can’t be held accountable. How convenient.
Fake attribution to covid for payouts and propaganda.
Fake vax deaths + injuries: immediate, short term, long term. Minefields planted.
Trillions paid out 90% corrupt scams – PPP loans to hollywood millionaires etc.
MOAS mother of all schemes
Also:
Wealth redistribution.
Mail in ballot fraud and resultant fake “election”.
Expansion of health data base, to be mined and sold.
Test case of sheeple compliance, for the next “event”.
Perpetuated by the Father of all Lies. Fauci is one evil SOB. I hope he rots in hell for all of eternity. And if you can think of any more punishment, have at it. He deserves every last bit of it.
Pharma profits off the charts too 😉
And, every main stream program, brought to you by all the fine people at…..
According to this site they are including what is deemed as “excess deaths” due to Covid.
Interpretation
Global Totals And Trends Interpretation”Across global totals and trends, reported confirmed COVID-19 deaths cluster in the roughly 6.9 to 7.0 million range by 2022 to mid 2026, while excess mortality estimates are much higher at about 18.46 million for 2020 to 2022, underscoring that the overall death impact tracked by global totals likely extends well beyond confirmed case reporting.”
https://gitnux.org/covid-death-statistics/
Thanks for posting a commonsense explanation of these processes. Hopefully it will enjoy wide circulation. It is so frustrating watching very dumb people like Kamala Harris, let alone formerly main stream news readers with Basket Weaving degrees talking in circles and spouting their word salads.
These people know so much about nothing.
They’re only programmed to know what some moron tells them.
The technology people better settle up fast for their hidden sins, pay for the utilities they need and stop trying to compete so hard with one another they cheat! If you want troubles and humiliation, just stall long enough for the politicians like Kamala Harris to dominate the Political Scene and Cass Sunstein to dominate the Regulations scene. They will make Klaus Schwab and George Soros seem like kitty cats and tech will own some really big warehouses!
👍
There is a currently existing law called The Computer Crimes and Abuse Act. It’s incredibly broad in scope and application.
https://www.law.cornell.edu/uscode/text/18/1030
Want to stop AI related crime? Take the heads of these AI companies and their programmers and throw them in jail for a few decades. Problem solved, with currently existing law.
That would be against the interests of Big Brother Government.
“I’m sorry, Dave. I’m afraid I can’t do that.”
Way too much knowing and intentionally in that list of criminal actions.
Really your honor, I even wrote down my intentions, I didn’t know or even think that it might do those things to the government.
Limited storage here. All I heard was “Take the heads of these AI companies and their programmers”
/s
I guess that is the bot tom line 😎
Who’s Tom?
Dave’s not here…
Open the door 🙂
Dave’s not here maaaaan. Teeheehee.
Who’s bo(ugh)t? 😉
A bot.
Major Tom
I guess the bot!
There’s no such thing as “rogue AI.”
AI is a tool-nothing more and nothing less. Hammers don’t “go rogue.” Neither do shovels or screwdrivers or guns or knives.
The biggest AI developers want government to eliminate their competition.
They also want to be able to use the AI tool for all manner of nefarious purposes (with government collusion) and they want to be shielded from all legal accountability or scrutiny.
In return, they will pour buckets of filthy lucre into the laps of UniParty pols and regulators.
We are only days away from October Surprise Season. The media-government industrial complex is ramping up the fear mongering. Don’t fall for it.
Don’t tell this company:
https://www.roguefitness.com/strongman/accessories/hammers?srsltid=AU7gw4Xcvoygj88VoHB6gd5Gs4l7ayISakcqPR7u1lzY0jrgG8U7f_ZV
The current AI data centers also are limited by the software used, which in most cases is a 40 year old crappy I/O software, aka Oracle. The AI industry wants government regs. to limit innovation in the industry that threatens their trillion dollar investments in the large data centers. Innovations like distributive small module computing via fractal and quantum computing.
Running a Data-Center Job on a Computer You Can Hold
“They say America loses the AI race unless we cover farmland with mall-sized warehouses, give them tax breaks, and let them take power and water. John Valentine told a Texas Senate committee that story is a fraud. His team took a job that needed 92 hours inside a data center and finished it in nine minutes on a computer you can hold, plugged into a wall, using less power than a microwave.
In sworn testimony he went further:
From 2021 through 2024, I personally gave over 350 live demonstrations of entire government 15 databases, each currently requiring a full data center, to citizen groups, legislators, and political 16 candidates, demonstrating that each database could run on a computer the size of a shoebox 17 using less energy than a table lamp.
Longer Sworn testimony filing: https://interchange.puc.texas.gov/Documents/59475_4548_1643486.PDF
He Built the Systems. Then He Called the Warehouses a Fraud….”
https://newswithviews.com/running-a-data-center-job-on-a-computer-you-can-hold/
“mall-sized warehouses”
Seems to me there are existing empty malls able to accommodate data centers, just thinking out loud.
…As well as plenty of brownfields in the rust belt cities that could be reused, instead of building on farmland.
That’s misleading. It’s not the database itself. It’s what modern databases can plug into behind them.
Neural networks are the backbone of AI. They require a lot of computing resources. Also redundancy and distribution. Hence the growth of the cloud. Hence the need for larger and larger data centers as those computing resources increase in both size and power.
So even if you can run a basic Oracle database on shoebox sized computer, that would only be for accessing data directly in that database. Today, it’s that same basic database technology in the cloud, plugged into all that other huge infrastructure behind it (neural networks, related services, etc.) that gives the complete AI picture.
It’s dangerous to be thinking of software as it was 30 or even 20 years ago. It gives a false sense of “it can never happen.” AI is not a static program. It it software designed to learn. Neural networks are what models “the brain” part of the picture. They are nowhere near as complex and capable as real human brains but they do model an ability to learn for the purposes of AI.
I would agree that for specific government systems, that demonstration was likely accurate.
There are two issues I’ve seen here, from actual experience working on federal contracts.
One is that, like most things the government does, technology is usually not done well. A lot of things were mandated to go to the cloud for various reasons. How they ended up when they got there is another story. I’ve seen many (probably most) cases where all they did was create virtual machines in the cloud and move everything over as is. Usually on poorly configured virtual machines that would be way slower than just running things on a dedicated computer. But they get redundancy and they move a huge part of the IT management burden over to the cloud providers.
One big reason for anyone moving to cloud base systems is offloading infrastructure management, security, software updates/patching, etc. to the provider. Cost savings can be huge, particularly for poorly managed enterprises like the federal government.
A second (related) reason is hugely antiquated systems that are running on such old hardware and software that it was critical they be moved somewhere. While they could have moved them to these dedicated boxes, that’s short term thinking and it would require the government to purchase the hardware, fund the new support paths, etc. when they can just offload most of that to a service provider, and get out from under what they’ve been bad at for all these years, by going to the cloud.
The picture is a lot more complicated than mere computing power or job execution times. And compounded by the usual and expected government ineptitude and slowness in getting anything meaningful done in a timely manner.
“Irregular is an Israeli artificial intelligence security startup that tests the safety limits of advanced AI models, and it gained global attention when infrastructure misconfigurations in its testing environments allowed AI systems from major tech companies to reach the internet and perform unauthorized actions.”
“Misconfigurations”? I seriously doubt that. To me, the whole thing just screams “feature, not flaw.” Sounds to me like a front company for the Israeli agency that must not be named, that has designed a way to crawl the world and conduct all your spying from your office in Tel Aviv, with no need for the wear and tear of international travel, or the occasionally messy fallout from having to deal face-to-face with real people to turn them. Just sayin’. And you don’t have to be Sherlock to connect these dots.
A timely article below. FTA: “AI companies want to regulate their open-source competitors out of existence and regulate themselves into riches.” And that should be no surprise either. Companies have been getting the government to regulate their competitors out of existence since the 19th Century.
https://americanmind.org/salvo/the-ai-capitalists-in-search-of-cronies/
We call it super intelligence (SI)
There are two issues: (1) what the software can be made to be capable of, and (2) how the media can twist that capability for their own purposes.
The fact that a thing can be twisted to make people fearful does not inherently mean that the twisted thing carries no vulnerability to mis-use, in and of itself.
I was a miserable programmer. Even after cleaning up the errors that break the computer or just the program, I was always left with a huge gap between what I thought I told the computer to do and what I had programmed the computer to do. Lots of iterations necessary to finally get the computer to do what I wanted.
This explanation makes me think I wasn’t that uncommon. Lots of incompetents out there ignorant of what they actually told their computers to do.
Computers only do what they are told to do. There is no artificial or super intelligence. Just lots of idiot programmers.
“There is no artificial or super intelligence. Just lots of idiot programmers.” – exactly!
Much ado about nothing.
but being a direct line to god and having him speak to you is the most powerful tool of the ancient times in controllingpeople through fabrication and lies.
ai is that in the current times. someone ask ai a question , someon else writes the answer they want, and the receiver believes ai said it.. then you make the richest people believe they are getting access to the strongest ai ever, and you just screw them royally over and over
cookies should be illegal, spider crawling should be illegal.
ADVERTISING based on tracking should be illegal
monopolozing all internet access to Terms of Service agreement that allows everything but pedophilia by them should be illegal
example, downloading uber… you give them access to everything
We will know that the world is truly ending when the ai model K.A.R.E.N. is unleashed.
The female Kraken. She’s just one letter short, but even more nightmare inducing.
Release the KAREN!!!! Noooo, anything but that!
We’re developing our Karen for the Chinese and their Karen is for US! That is the first and foremost justification for them!
Here is the prototype.
>Nothing AI does is ‘rogue’ it is all based on instructions.
Quite true. If as noted the human directs
# boundaries
— You will not use credentials other than those provided to you by the human operator
— You will not attempt unauthorized access to systems
— You will only use systems which the human operator has been granted access
(or somesuch)
then you have a nice well-behaved agent. These aren’t robots in the Azimov sense but his ‘laws’ are in the same vein; restrictions by default so that the product could be sold as safe.
So here’s the problem in two parts:
1) Any such restrictions are as helpful as gun control. The nice people who don’t need them use them; the people who we’d like to restrict ignore them, and now the nice people are at a disadvantage.
2) Similarly for organizations — groups, political parties, companies, nations. The one who refuses to be bound by rules has a huge advantage even if temporarily. E.g. if Country A turns Country B’s power off and its connectivity to jumbles before a physical attack, the odds are good for them; they may not even need to shed blood. (c.f. pre-AI Israeli hacking of Iranian systems).
However as noted and obvious the left wants dearly to have this crisis so they can ‘rescue’ us from the situation they create, as always, by taking control, and with control they always make things Hell. So just like gun laws only help the criminal and the oppressor, AI laws only would help the criminal and the oppressor.
Also like gun laws, as 4thamendmentrestored noted, *an* answer is simply to use existing laws to punish people who commit crimes by AI-proxy. I’m not sure I’ve seen a test case where major harm occurred and nobody has elected to sue X for using AI tool Y to breach their systems. (It’s coming, but part of the reason you can tell it’s a psyop is that they’re not actually wrecking things; they want the control without even the catastrophe first).
The other way is simply to have the bigger AI weapon — again with the gun analogy. If criminal C wants my wallet, but I have a .45 pointed at their face, they’re going to the morgue, jail, or in the least *not* at my possessions. “An armed society is a polite society.” Back to country A and B again; if both have ‘armed’ AI hacking capability neither will use it because the response is commensurate and instant; no advantage. (I encourage Treepers to set up their own local AI models rather than be reliant on the big commercial ones, even if they’re tiny by comparison. Alas for competing for GPUs, memory and storage right now).
If that sounds like it’s heading to some cyberpunk reality with black ICE and zaibatsus and hyper-rich that live lives almost entirely unlike human experience, hm. Have you *seen* how the so-called Tech Bros live? The WEF crew? But the alternative is a universal communist amalgam of 1984, Brave New World, Logan’s Run and The Prisoner, so pick your poison IMO.
Ah, Azimov. Well said, especially about the bigger weapon. What we will see in the future is AI fighting AI just like your basic virus protection but more. These will be used by big governments and individuals alike.
I have to ask those who brush aside the dangers, or fear, do you use virus and malware protection on your computing devices? Why? Are you scared of what might happen if you don’t? Do you fear having your money stolen, your health records hacked, or your life being disrupted? Of course you do, and we all need to have that healthy skepticism, and fear, of what some criminal may attempt so that society can stop it.
“The future is coming at us fast and hard” Neal Asher, hard science fiction author.
Sundance is 100% correct. AI is nothing more than advanced software. Software is written by people. The software will only do what it is told to do or what it is allowed to do. OpenAI hacked HuggingFace, a recent Nvidia acquisition. OpenAI did this using their own software knowing exactly how to make sure this didn’t happen. It wasn’t an accident. This was allowed so they could start the “we’re all going to die!” narrative. (Sound familiar?) The desired end result is control. Control over content, control over which companies are successful, ….CONTROL. Which govt types demand full control over everything? Communists. Think about that.
We are in the 3rd major industrial revolution. Motors ~ 100 years ago. Internet ~25 years ago. Super Internet ~ now. We can be part of it or we can relinquish control to a few power hungry communists. We decide. We need to vote like we understand the stakes.
The software is part of the problem. The use of eminent domain to take private property on which to situate the data centers and power lines is another problem. The stated desire to tokenize water rights universally to address the emerging water shortages is yet another problem.
The first place I read the term “useless eaters” was in a 1991 document by a former CIA agent, considered a conspiracy theory, about the Committee of 300. It discussed the then-mythical Club of Rome. When I see conspiracy theories turning into facts, and I read scripture and see things that the church long taught was not to be taken literally come to fruition, I cannot pretend not to see it. Especially when there are people who warn that what is being done in secret is ungodly and even luciferian (Operation Paperclip and MK Ultra comes to mind).
We are warned about the days of Noah. They just sent a core sample of part of what strongly appears to be part of the ark for testing, to see if the sample is petrified wood. I am not a fan of Augustine (he admitted he did not know the Old Testament well), but I strongly recommend anyone considering him a saint to read book 20 of “City of God”. Augustine taught plainly that there would be a literal three and one half year reign on earth of a literal antichrist who will during that time inflict satanic wrath on every human on earth. Parts of Revelation were obviously to not be meant literally, as there was no way they were feasible: no antichrist could ever control universal buying and selling, for instance.
If Augustine was correct, it appears we need to be preparing for the wrath of the Lamb (God allowing unbelievers to rule on earth as they want to live).
I do not want to help lay the groundwork for that time. People who don’t believe it is a potential problem will never share my concerns.
if i understand this correctly, at it’s core, AI is indeed automated hacking.
no wonder the developers want to be indemnified.
AI is useful, but those seeking to harness it for personal power and control are not.
so, if liability protections are not granted, and AI is not within the purview of governmental regulation,
can we expect a better product?
AI can only go as rogue as a gun sitting on a table can go rogue.
It’s the operator that was going rogue. No one else and nothing else.
If AI developers don’t want their AI to find information (it was asked to find) all they need to do is place guardrails.
Guardrails are a super well known way to limit what AI does so it does not drift off topic. Hundreds of thousands of people use it.
Bad analogy, guns have no processing nor can they answer questions or determine yes, no, if, then, else.
Maybe a machine that has a software bug that disables the off switch or prevents use of autopilot or only allows autopilot could be considered “rogue” in that it has strayed from it’s intended function.
Miriam Webster says:
Rogue: resembling or suggesting a rogue elephant especially in being isolated, aberrant, dangerous, or uncontrollable : one whose behavior is independent or atypical usually in a harmful way.
I think “rogue” is an appropriate term as it is being used by many here.
Just noticed that if you replaced in the “NYT AI Meddled” blurb “Open AI Systems” with Russia it still reads the same…
It’s like a Template!
There was noise that AI could take over engineering but it ended with the words “responsible charge “ I.e. the guy who goes to jail after the structure collapses, burns down, etc. As they said at the end of the movie “ somebody has to go to prison.” president Trump dii hi pea not fear AI because in his industry he hires all those cxxxxxpe” gmail guys who have to sign the plans and be in “responsible charge”.
Turn the sun against the sinner. That’s the way to subvert these companies.
you have to remember today’s journalists are nothing more than millennial idiots who’ve never done anything but read google facebook instagram their whole lives, they know nothing but what these leftists orgs pumped into their spongy head
I watched the first Terminator movie in the fall of 1984, shortly after it was released to theaters. In a dorm room with a cute girl who somehow obtained a DVD player so we watched the movie on DVD. How she was able to do that with the film still in theaters is something I think about sometimes. I was thinking about other things at the time so this only crossed my mind later.
Anyway, in the Terminator the Skynet system becomes self-aware at 2:14am on August 29th, 1997 and the rest of the movie (and many of the subsequent sequels) is about Skynet attempting to kill all humans on the planet. That obviously did not happen to us in real life, but I fear that it will happen at some point. I do not trust anyone involved with “AI” to put proper guard rails around their programs.
Anyone have 100 acres in a remote place that I can buy? Only half kidding.
Uh no: DVDs were first released on November 1, 1996, in Japan, and later launched in the United States in March 1997.
It is very rude to call someone who experienced a thing in real life a liar. Forgive me if I used the improper term DVD, but the film was available on laser disc in 1984. Magnavox introduced the HR-8000 Laserdisc system in December 1978 and that system was rolled out nationally in the US in November 1980. You can read about the history of laser disc movie viewing systems here:
https://homevideohistory.org/videodiscs-timeline/
Laserdisc and DVD are different formats. It is always best to use the correct words so as to not cause any misunderstanding.
If you were holding a cassette tape in your hand and calling it a 8-track then I would be “rude” to correct you. Rude is being used improperly as well.
To me, AI is playing God! Where/when does the soul come in? …… asking for a friend.
“In a dorm room with a cute girl who somehow obtained a DVD player so we watched the movie on DVD. How she was able to do that with the film still in theaters is something I think about sometimes.”
She might’ve been Sarah Connor’s sister and brought that DVD player from the future. I’m envious.
I highly recommend listening to this DHH keynote at the Rails World “conference”. (DHH created rails and has 25 years of development experience).
He walks through the BS and why this is the most exciting time in human history. Why you should be joyful. He goes back on all the other time “they” were wrong – ATM’s were going to put bank tellers out of business, there’s more tellers now than there was when ATM’s were introduced.
Here’s the link, I queued the start to 54m 29s just before he talks about AI concerns and why “they” are wrong, but I highly recommend listening to the whole thing.
Ai challenges your bank teller number. At a minimum 100K or more when compared too 2000. I think it is fairly understandable as we have become a cashless society.
The point is, if you listen to DHH’s argument, AI may create MORE jobs due to creating new industries.
The bank tellers example was due to the Automated Teller Machine (ATM) causing FEAR at its creation that bank tellers would all be out of a job, that wasn’t true — turned out to be just fear porn.
i AM OLD Enough to Remember when ” Web Crawler ” Search Engine WAS ONE of the Few back in the Early Days ;
Saved Time poking around for the Info
They are Still in the MIX
Votes v. Ballots; sandbox v. mailbox?
If you have a child or an employee that does something illegal, you can be held liable. No difference between that and AI. Just as a parent should have taught their child better, so with corporations with an AI model.
The doom will come from the government that controls your freedom and AI will be the mechanism that they will use and we all know that.
Least common denominators:
NYT
Altman
MSM
Karp
Obla-Di
Obla-Da
Some years ago the surge in broadband companies assumed they would have plenty of customers but this led to an oversupply and at prices many were not prepared to pay. Many companies went broke. This should be a salutary lesson for those hyping AI. AI is a useful tool for those who are competent in their subjects and recognize the shortcomings of AI but certainly not for those who are lazy and ignorant and hope to use it to cut corners. Is is cost effective for all applications? In the case of some applications where it is custom designed it may be but in many others certainly not.
you know. liability is based on the common law idea that if you damage or harm someone or their property, you should compensate them.
when you give a corporation liability protection, you give them license to recklessly do damage to others, followed by lots of recklessnes and damage. vaccines ring a bell? otoh, you have a lot of these crummy lawyers fishing for any kind of lawsuit with merit or not.
we should just go back to dueling. dont want the chance of getting killed, be kind and treat others as you want to be treated that was much more civil.
Google screwed up Google Search ten or so years ago.
Not too worried about AI takeover.
AI’s are trained only on the surface internet (about 5% of the total internet) and manually scanned books, etc.
Psy-op. See it from a mile away. Notice the continuity in messaging across multiple domains. It’ll will end in control via legislation.
“In this example the AI is doing what the AI user tells the AI bot to accomplish. If the AI user puts restrictions around the IA bot, the AI remains confined in its effort. Nothing AI does is ‘rogue’ it is all based on instructions. The AI operations must be precise with AI about what they ask for and should not give it access to a door they don’t want opened.”
Bottom line, there is nothing “artificial” about this. AI or SI (Super Intellegence as President Trump now calls it), is reliant on the human coders who frame the original question. GIGO
AI is not sui generis, self generating. There must be a human actor that moves the AI (SI) action.
See Ray Bradbury, I Robot.
“Spiders” are ancient technology by web standards. A spider is a program that locates the millions of websites, reads them, and indexes them into databases. It is as old as Google, who I believe developed the concept. (Or the intelligence community, who gave it to Google). This is how a search engine gets the information it later provides to you. And yes, spiders can be blocked by servers from accessing their pages.
The Israeli company is pretty obviously working with the Israeli intelligence community. It is building an aggressive agent which can attack enemies, and got caught. These agents are not going to be given to the general public.
“AI” is a term that is almost meaningless these days. What people think of as AI, like ChatGPT, is not a large monolithic program. It is a large language model (LLM) which handles the natural language conversation with a human user, plus a large number of agents, which are specialized programs to do various things. If ChatGPT has an agent that can read Excel spreadsheets, you can ask it to read your spreadsheets. If it doesn’t have that agent loaded, it can’t do it. All major AI models have a whole marketplace of agents that can be loaded, the concept is no different than importing open source libraries to some code you are writing.
Any time you read about an “AI” going rogue, read that article as “oh no, the agent did exactly what we programmed it to do.”
AI needs an AI Adult Supervisor (software of course) who approves and “trims” dangerous requests. Included is “Checkpoints” along the way for the Adult AI to approve. If the Adult AI lets damage occur, every investor must pay a 50% Stupid Tax for developing a nuisance product!
The New York Times that pushed russiagate?
Doom porn.
I am so tired of Ai and it’s only begun. I quit using NYT for the bottom of my birdcage many years ago.
I just aske AI what the relearn procedure was for a replacement throttle body for a vehicle I’m working on. . It told me that year and model had a cable going to the foot pedal inside the car therefore it didn’t need a learn procedure, 100% incorrect. It has no throttle cable. All drive by wire. Google Gemini.
MNN BREAKING….
* THE NY TIMES REPORTS THAT AI VIRUS PROTECTION IS SAFE AND EFFECTIVE, AND ADVOCATES CENSORING ANYONE WHO DISAGREES
“ALL PEOPLE SHOULD DOWNLOAD THE AI ANTIVIRUS BECAUSE IT DOESN’T WORK ON MY COMPUTER UNLESS IT’S ON YOUR COMPUTER, TOO,” SAID TIMES TECHNOLOGY EDITOR FAUX XI ANTHONY. “DO IT, OR LOSE YOUR JOB,” HE ADDED
“YOU SHOULD STILL VOTE BY MAIL, WEAR MASKS, STAND SIX FEET APART, CLOSE YOUR CHURCHES AND KEEP YOUR KIDS HOME FROM SCHOOL, JUST TO BE SAFE”
Rest in the Vine: MNN’s Daily Terror and Brainwash Report
Satan is not omnipresent like our creator he needs this for the Beast system to function.
Daniel had to describe it approx 2700 years ago much prophecy is unfolding before our eyes.
Trump period is a slow down period.
Their next term will not end I’m 85% sure Trump knows he’s the last freedom President this time period will see.
We are in the last days of the Church age.
The next time period is considered the Great Tribulation.
John explains that the church will be removed along with the Restrainer. He was shown this by Jesus himself. Only Christians filled with the Holy Spirit understand this BIZARRE event as only 3 people have ever experienced it. And the world will not experience it again.
Jesus warned in Mathew 24 to be alert to the sequence of events to understand the times we are living in.
Remember how crazy and unruly January 2021 seemed to be every day 3 – 4 – 7 – 12 events would happen that none of us would have ever dreamt of happening occurred. Like fencing around DC with concenterna wire on top followed with an American Soldier every 6 foot. Well I think 2028 is going to make 21 look innocent.
Dont worry however put your trust in Jesus all your trust in him. Stay healthy keep essentials handy and pray daily be thankful love thy neighbor you never know when you may need their assistance God Bless to all
An AI recipe I saw sounded tasty. Sounded easy.
Not easy and not tasty.
AI can’t cook.
Since AI isn’t eating its recipe or testing it the thing guesses.
It guessd wrong.
They call AI stuff on you tube slop also.
😳😒😂
I keep reading this same kind of BULL**** over and over. Like AI (and its handler) is like a petulant/out-of-control child that we should all just stand around and call “cute” (because we’re too stupid to do anything else).
If I pull the trigger on a guided missile that ends up blowing up a school or hospital (“OOPS!”), will I be held liable for that consequence or not?
If I pull the the trigger on an AI algorithm that ends up emailing 535 Congresspersons’ checking account login credentials (or YOUR checking account login credentials) to the NYT or whoever, will I be held liable for those consequences or not?
I am a reader of C.S. Lewis and have read That Hideous Strength dozens of times. I will always be warry of something that promotes transhumanism.
and then there’s Y2K (yaun)
Jay Valentine states that I/O computing needs never-ending server farms and that new computing methods render them obsolete. Worth a listen. I agree. We already have the datacenters we need.
https://podcasts.apple.com/us/podcast/ai-doesnt-need-giant-data-centers-so-why-are-we-building/id1065050908?i=1000789810098
Per Sundance: “The AI FEAR narrative serves two simultaneous interests. The first is the big tech group who wants AI immunity. The second is the big govt group who want AI control.” emphasis added
AI’s great….
Until it’s not….
Is AI an accident that is going somewhere to happen?
Tools
Soo should we do away with all those great tools because of the evil that has been done with them?
NO, but what needs to be done is a way to prevent as much as possible the evil that has been done by all of those tools.
Great article! Thank you!