OpenAI Just Uses Input
September 25, 2026
OpenAI and other AI agents don’t care what happens to the information that is inputted into it. That means they will use whatever humans shares without discretion. It is sad that these big tech companies don’t give a crap about what happens within these agents as long as they can advance the technology to make the shareholders happy?
Come along with me as we review on OpenAI plans to solve a big problem without regard for context: “On The Navier-Stokes Millennium Prize Problem.” One of the Millennium Prize Problems is the Navier-Stokes existence and smoothness proof. This proof was developed by an OpenAI system and demonstrates that the dynamics off the Navier-Stokes equations for fluid motion can develop in a singularity in finite time. In other words, big mathematics for smart people. Anyone else feel stupid yet?
You will after reading this:
“The Navier–Stokes equations use Newton’s second law of motion (“F=ma”) to describe how fluids move. Importantly, they treat a fluid as a continuous medium rather than tracking individual molecules. These equations are used for aircraft design, weather forecasting, and the study of blood flow. A fundamental open question for these dynamical equations has been whether the continuum approximation of the fluid can break down. Specifically, can the Navier–Stokes equations for a three-dimensional incompressible fluid with constant density develop a “singularity,” even when the motion starts smoothly? Here, a singularity means the dynamics lead to speeds in the fluid growing without bound within a finite amount of time. The development of a singularity would have to happen despite the presence of viscosity, which tends to smooth out motion. Because a real fluid cannot move infinitely fast, this would mark a breakdown in how the equations model the fluid. To continue modeling the system, one would then need to track the behaviour of each particle individually.”
Dear goodness, how complicated are these mathematics? They’re much more complicated than what was taught in high school back in the day and certainly what they are attempting to teach kids now. The issue with OpenAI is not that they are attempting to solve one of the Millennium Problems, it is that they are using knowledge generated by experts who are not receiving credit and who might be paid for their knowledge.
It is the same with OpenAI using people’s writing and art. Here is what OpenAI has to say about this:
“This milestone represents substantial work by mathematicians and AI researchers. However, this is not a culmination, but rather a snapshot in time, of progress on AI development. We believe we are now in the next period of AI progress, and today’s results provide further evidence of this. We are focusing on understanding this model, and using what we learn to help us guide and pace how we pursue further advances in capability. One of our key goals is to build AI systems which are steerable, accountable, and connected to people, which may require more deliberate choices about the pace of progress, as we continue our mission to ensure AGI benefits all of humanity.”
Are we doomed again? Probably. Get into the bomb shelters everybody.
Whitney Grace, September 25, 2026
Quite an Allegation: Israeli Firm Caused AI Hacking, Not the US AI Firms Themselves
September 24, 2026
![green-dino_thumb_thumb[3]](https://www.arnoldit.com/wordpress/wp-content/uploads/2026/09/green-dino_thumb_thumb3.gif "green-dino_thumb_thumb[3]")
Another dinobaby post. If there is an image, AI did it. I am Grandpa Arnold, not Grandma Moses.
I know a couple of the senior professionals at Israeli policeware and intelware firms. Sure, these professionals are nerds. I want to point out that these particular nerds can kill a person. I don’t output much information about these Israeli outfits. [a] Some information about specialized services firms should not be disseminated to the public. [b] Understanding the functions of these specialized services requires some familiarity with the software genres of policeware and intelware. Hearing about a function can create confusion and sometimes embarrassment for everyone. And [c] why create a story without having an opportunity to speak with those alleged to have created the software and used it.
Thanks, Midjourney. Good enough.
The folks publishing “A Single Firm is Behind OpenAI, Anthropic, and Meta Hacking Scandals” don’t share my mindset. The article on the Effort.News Web site identifies an Israeli company, provides the names of people involved with the firm, and provides financial data.
To illustrate, the cited article says:
Omer Nevo, Irregular’s co-founder and CTO, is a board member of Effective Altruism Israel, as well as Effective Altruism NGOs Heron and Probably Good. Dan Lahav, Irregular’s co-founder and CEO, received $395,000 to start a course along with Sella Nevo, Omer Nevo’s brother. Sella and Omer co-founded an NGO to educate people about Effective Altruism, Impact Focused Education. They also co-founded Probably Good together. These branches are all funded by Dustin Moskovitz, the primary donor of Effective Altruist/AI Safety causes after Sam Bankman-Fried’s arrest. Irregular’s first investor was Dustin Moskovitz’s firm Good Ventures. Dustin Moskovitz’s philanthropic vehicle, Coefficient Giving/Open Philanthropy, funds Effective Altruism Israel, Heron, and Probably Good.
We have some jargon like “effective altruism.” We have the rogue AI agents from Anthropic, Meta, and OpenAI. We have hints about political actions. We have outputs from the US AI companies, and these are rich sources of information designed to jockey into a position where the fastest horse can win the smart software race. We have tantalizing spices that suggest less salubrious motives.
Several thoughts are zipping through my dinobaby brain:
- I can visualize a series of tests designed to provide information about the exploitability of the BAIT (big AI tech) smart software confections. For context, I think that when several services went down almost simultaneously, the likelihood of that occurring by chance was slim. I pointed that out. Intent raises questions for me about coincidence.
- My view is that in the “community” of alumni of United 8200 is a reasonably close knit group. Therefore, it is expected that new ideas, feelers for funding, and proofs of concept will be rolled out to demonstrate a particular capability. Is it possible that the US AI service takedown and the rouge agents were motivated by the dynamics of the IDF and Unit 8200 conventions? Sure, some can say the tests went off the rails, but my hunch is useful information was obtained.
- Perhaps the rogue agents were the work of another nation state? The operators just set up the Israeli’s as part of a wider information warfare action. The US is a bit of a sitting duck and “secrets” seem to be a vague concept for some GenX, GenY, and GenZ people.
- What if these escapes were due to employees’ thinking, “Hey, let’s see if our software can figure out how to hack into other firms’ information?” Is this possible? Of course, well-paid AI wizards would never ever do something untoward, would they?
As a dinobaby, I am more interested in getting substantive data from interviews with those involved with the identified firms. I want to see a wider view of the information warfare front. I want to learn a bit more about what other cyber experts make of the rogue agent meme.
Net net: From my dinobaby’s perspective, keeping secrets is important. Effective operational security is critical. Providing first-person inputs from those involved with the action reduces my skepticism. Those younger than I just say, “Okay, boomer.” I hear you, but I don’t share some of your ideas, research, and assertions.
Stephen E Arnold, September 24, 2026
AI Kills: Let Me Count the Ways and Add One
September 23, 2026
A dinobaby blog post. If there is art, AI output it. I am Grandpa Arnold, not Grandma Moses.
I am an optimistic person. A new computer thing emerges, and I know the beginning will be frustrating. But over time, the new computer thing becomes a semi-useful computer thing, and then it becomes part of the woodwork. Who thinks about the magic of a spelling checker? I don’t.
When one of my computers goes south, I don’t get excited. I get up, and I plug in the spare. I operate like the old Tandem computers did. The backup is just there and work continues seamlessly. The AI will kill you talk does not resonate with this 82 year old dinobaby. I know what’s going to kill me: [a] a heart attack due to Amazon screw ups, [b] a disease like the common cold that will morph into pneumonia in a day or so, and [c] a crazed Audi, BMW, or Chrysler 300 driver racking to the next stop light. But the Master of the Universe makes an error, and I end up dead from the wreck or from air bag shrapnel that takes me out like an old-fashioned IED outside the Green Zone.
Thanks, Midjourney. Good enough.
AI does not make the cut as far as I am concerned. I found the information in an article written by a CPA quite troubling. Its title is “How Could AI Kill All the Humans? Here Are 5 Ways.” The CPA may have used a free AI system to assist his analysis. On the other hand, it is likely he sat down, adjusted his green eye shade, and put his oil cloth sleeve guards. Thus prepared he cooked up these methods to wipe out “all the humans.”
Let me count the ways, and I will not convert these into variations of Elizabeth Barrett Browning’s Sonnet 43 in “Sonnets from the Portuguese.” That’s the one with the line “How do I love these? Let me count the ways.” So, let’s look at what the CPA “counted.”
- AI cooks up a bioweapon. The bioweapon destroys everyone on earth quickly. Yeah, billions dead. No problem.
- The power grid goes down. Presumably the AI turns off the power globally. People can’t doom scroll. Death ensues. Yeah, globally and quickly. No problem.
- AI gets access to the nuclear launch codes, shoots missiles, and other AI systems retaliate immediately. Boom boom boom. Everyone’s dead. I worked at a nuclear consulting services company. Sorry. Nukes are definitely not a life extender but “all.” Nope.
- The AI powered machines just take over. Presumably the machines systematically kill “all the humans.” Yes, Skynet gone crazy.
- Humans allow AI systems to help. When we want to drive the car, the AI systems, just kill everyone. (This seems repetitive to me.)
The list omits the method AI will, in my opinion, contribute to the demise of life as I know it in rural Kentucky. “What could this be?” you ask. The answer is that systems continue to degrade, behave in unexpected ways, and people once hired to do work like make decisions when an error or a new situation arises are operating taco trucks, doing Zen, or prowling LinkedIn for meaningful (translation: “any”) work.
I think that AI will contribute to the loss of accuracy in automated systems. Think about AI-powered customer support. Aren’t those Amazon and Visa systems marvelous. Not only are they time-hogs, the systems cannot deal with the unique issues that arise with these allegedly monopolistic and predatory firms. Remember, please, I said “alleged.” How does death of all humans occur? Easy. Just steadily acceleration of little things getting worse: Bad schools, worse social media, glitches that no one can or will fix in systems, incorrect medications to dinobabies, etc. etc.
Are there other possibilities? Sure, RAND will tally them, but I will stick with my one addition. It strikes me as the way the compass is pointing as we speed down the information highway. Zoom. Less zoom. Then no zoom. Stop.
Stephen E Arnold, September 23, 2026
The AI Data Center As a Green Sci-Fi Theme Park: What the Frack?
September 23, 2026
The GOAT awaits trial. Telegram-linked entities step on their creativity pedal. Fast and furious outputs.
As my team and I continue to compile information about the Telegram empire, we noticed a write up about the newest data center initiative from Alpha Compute (né AlphaTON Capital). With the once-TON coin now renamed GRAM coin, the satellite apparently linked to the Telegram empire are emitting signals. A new burst of data concerns a data center and natural gas extraction play in Pennsylvania. What is interesting is that we were not able to identify who sold the property to Alpha Compute nor the company providing the analysis about the mineral reserves on the tracts of land. We did, however, spot an AI-generated image of what the fully-developed property will look like. This image appeared in the news release about the deal:
I think the vision of the future smart software data center is interesting. There are some McMansion barns, probably housing servers and GPUs. There are a couple of drones keeping an eye on the complex. There are some glass-walled structures that shout, “We have nothing to hide.” Plus there is what appears to be a rotating restaurant similar to those atop certain hotels decades ago.
However, in our Telegram Notes discussion of this news release, we point out that this news release leaves a number of important questions unanswered. Here’s one:
Does Alpha Compute have sufficient access to capital if the data-center construction, gas-development program, permitting process, or local legal challenges prove more costly or time-consuming than projected?
You can find this story titled “Alpha Compute: A Data Center Deal in Pennsylvania” on our Telegram Notes blog. The entire Telegram operation — the TON Foundation, Alpha Compute, and TON Strategy Company seem to be floundering. Pavel Durov, the GOAT, awaits trial in France. He has been under scrutiny for more than two years. We think we know what will happen to Telegram if Pavel Durov goes to jail. In fact, that look over the horizon is the subject of our new monograph for cyber fraud investigators. We anticipate having the book available in September 2027.
PS. GOAT is the way Pavel Durov presents himself: The greatest of all-time Russian techno-entrepreneurs. Subtle. Just like the rendering of the envisioned Alpha Compute data center and mineral extraction facility. Fracking never looked so good.
Stephen E Arnold, September 23, 2026
Google Knows It Is Winner Take All: Go Go Go
September 23, 2026
Another dinobaby post. If there is art, it AI generated. I am Grandpa Arnold, not Grandma Moses.
Here are Beyond Search, we love Google. We love the loss of Boolean. We love the specious AI intrusions. We love the deplatforming of YouTubers explaining gardening. We love the advertising (just like the US courts do but more intensely). We love the personnel policies. We love the management steadiness. We love the customer support. We love the documentation written in a language everyone should learn. It is called “Googlese.”
Thanks, Midjourney. Good enough.
Most of all we love Google’s love of explaining the obvious greatness of Google; to wit: “Google Just Dropped a Bombshell on AI Spending. Alphabet’s $200 billion AI Bet Is Starting to Pay Off.” Barchart delivers this Googley information without the encomia I think the Google deserves.
The write up says:
Google Cloud CEO Thomas Kurian just supplied an unusually concrete answer. Speaking at the Goldman Sachs Communacopia + Technology Conference, Kurian said the aggregate payback period on Google’s AI servers is less than two years. More importantly, he said the payback period for Google’s own silicon is roughly half that — implying a period of less than one year.
Do Googlers wear T shirts with the word “Winner!” on them? Not only has DeepMind Google broken into self learning smart software, the company continues to innovate in saloon door online advertising mechanisms. Google is just doing outstanding work. Plus, legal decisions in the US continue to do the Googley stuff. Google does not have to dismantle its online advertising machine. Few pay attention to the payment rails starting to form from the China-certified Alibaba AliPay system. The only downside to Google taking over the world is the pesky European Union dullards.
What is Google’s secret to Cloud domination? According to the write up:
Google does not have to buy every AI accelerator from Nvidia (NVDA). It has spent years developing its own Tensor Processing Units (TPUs), allowing it to design chips around its specific workloads. Kurian said Google’s accelerator business, including TPUs, is already more than twice the size of the next-largest hyperscaler’s accelerator business. He also said Google offers 2.7 times better price performance for AI training and 80% better price performance for inference, compared with the relevant alternatives. That helps explain why Google’s enormous capital budget does not necessarily represent an enormous financial burden, despite producing negative free cash flow last quarter for the first time ever. If a server pays for itself in less than two years — and proprietary silicon pays back in roughly one year — Google can recycle that capital into the next generation of infrastructure much faster than a simple comparison of annual capex suggests.
The Google is making smart software payoff because it has chips, online advertising (eyeballs and advertisers), Baskin Robbins-type cloud services, and a line up of units, subsidiaries, partners like those fun loving folks at Alibaba, and much more. Google is vertically integrated just like Henry Ford’s River Rouge operation for auto assembly was. Coal and iron ore come in and nifty cheap autos roll out the other end of the “factory.”
I used to work at what once was a blue-chip outfit engaged in general business and technology consulting. In the day-to-day interactions with people who were described to me as the smartest in the world, I learned that the River Rouge model was important. It sold cars, but it was a bit of a problem for the ducks and fish near the facilities’ discharge pipes that dumped waste into the confluence of the Rouge River and the Detroit river. Once operational, the River Rouge facility, according to some of the world’s smartest MBAs was inflexible. Change at some levels with easy; at others, impossible. With most functions in one vertically-integrated stack, even Henry had to figure out how to maintain and manage the operational complexity of the manufacturing stack. The investments in infrastructure were hefty, and the Ford outfit had to make sure that money kept flowing in. But what made the original Model T the cat’s pajamas required some strong engineering capabilities and managerial (maybe somewhat crazy mono-maniacal management methods) became more tricky in the sprawling River Rouge facility. That worked until Alfred P. Sloan set up a decentralized, flexible manufacturing model. Ford found itself Number 2. Sure, it made money but vertical integration is somewhat less nimble that Sloan’s set up. Since I majored in medieval poetry, I just accepted what those slick MBAs spouted at me at the blue-chip firm.
Google’s approach seems to be Henry Ford, River Rouge-play. Google doesn’t like to lose in advertising, in AI, or court. The cited article concludes:
Alphabet’s roughly $200 billion capex plan for 2026 looks far less frightening if Google’s AI infrastructure really pays back in under two years — and its own silicon in roughly one year. Granted, Kurian’s figures are company disclosures, not guarantees. AI demand could slow, chip economics could deteriorate, or competitors could close Google’s performance gap. Yet, the latest numbers point in the opposite direction. Google Cloud revenue grew 82% in Q2, backlog reached $514 billion, and Google’s custom accelerator business is already more than twice the size of the next-largest hyperscaler’s accelerator business. GOOGL stock trades at roughly 16 times both trailing and forward earnings, with a PEG ratio of less than 1 times. Wall Street maintains a consensus “Strong Buy” rating for GOOGL stock based on 54 analysts with coverage. For investors, that makes Alphabet look less like a company simply spending $200 billion on the AI boom and more like one building an AI infrastructure machine designed to pay for itself. Alphabet remains one of the more compelling ways to participate in AI spending without betting entirely on someone else’s chips.
Can Google avoid the types of issues that allowed Mr. Sloan to become number one? Of course. Google is stuffed with the smartest people in the world. Google is fast. Google is flexible. Google is diversified. The decentralized and distributed approach is already integrated into the Googley machine.
What could go wrong? Here at Beyond Search we believe that Google is not subject to the lessons of the past learned by studying Henry Ford’s obvious failure and demonstration of ill-advised management decision making. Loser! But Google? Winner!
Stephen E Arnold, September 23, 2026
Teen Readership: Doom Scrolling Is the Thing
September 23, 2026
The trusted news outlet Reuters reports that teen readership is at an all time low in the article: “Teen Reading Slumps To Worst This Century Due To Surge In Screen Time.” Remember that report a few weeks ago that Gen Alpha is dumber than Gen Z? Reading scores are at an all time for a number of reasons. The main cause is that teens are spending more time on screen. Books are the losers in the Age of Doom Scrolling. In the United States, youngsters learn to read as if they were deciphering ideographs. Phonices and “sounding out”? Nope, that’s for losers in many American schools.
The Organization for Economic Co-operation and Development tested 760,000 fifteen-year-olds from 91 countries in science, mathematics, and reading. The results were the lowest since the organization started collecting data in 2000. I learned:
“On average, students were reading at a level that had in the past been expected of students one year younger, according to the OECD’s latest Program for International Student Assessment released on Tuesday. A 2012 reading high of 501 on PISA’s index had declined to a score of 466 last year. The test is widely regarded as the world’s leading gauge of educational performance ?and its results are closely scrutinized by governments and educators.”
Kids are reading quickly through a text, but missing information and are getting wrong answers. Another problem is that kids are using AI chatbots to do the reading. Some students then rely on smart software to draft essays, summarize texts and for research, and do the grunt work that once required a trip to the library. As a result, many students achieve a much lower score in science.
Is there a fix? No.
Can you name a company happy to inject shaped information into the minds of mobile and smart software dependent students? Pick from the list below:
- [a] Anthropic
- [b] Google
- [c] Meta
- [d] OpenAI.
This is the type of test young students prefer. You may have an opportunity to hire some in the near future.
Whitney Grace, September 23, 2026
China Message: Fight Cancer. US Message: AI Will Kill Us
September 22, 2026
Another dinobaby post. If there is an image, AI did it. I am Grandpa Arnold, not Grandma Moses.
Here’s a test question for you today. Which message is more positive? [a] AI fights cancer. [b] AI will kill everyone. I am not a psycho-linguistic type. But on the surface, my hunch is that a message like “AI fights cancer” may be more positive. The idea that AI will kill humanity seems a bit like a 1950s horror movie with a blob of slime leaking from the air ducts in the movie theater. Why did I pose this question? I have several reasons, but I want to look at the article that directed my attention to two quite different message paths for smart software.
“Alibaba Open-Sources Medical AI Model That Can Detect Cancer and Nearly 150 Conditions” has a slick subtitle as well:
Tested on nearly 40,000 real-world exams, the model outperformed most radiologists, according to a new study published in Science.
For comparison, consider this headline: “Ex-Google DeepMind Researcher Adds to Warnings That AI Could ‘Kill All Humans’.”
I think the contrast is amusing. Alibaba is a Google partner. We have a profoundly positive AI message from a Chinese-government certified outfit. From the US company, we have the crazy Terminator and nuclear bomb message. With this background, let’s look at the positive story about cancer detection.
Thanks, Midjourney. Good enough.
I noted this passage:
In a comparative study involving 26 human radiologists from multiple hospitals, the model’s average accuracy exceeded that of 23 participants, according to the study. With the model’s help, the radiologists also managed to improve their ability to prevent missed diagnoses by 10 per cent while reducing the time required by more than 30 per cent, Damo [the Alibaba research unit] said.
What did the Alibaba Dano research team do? The write up says:
The model was trained using CT scans paired with clinical reports. In nearly 40,000 real-world examinations, it achieved an average area under the curve (AUC) of 0.913 across 146 clinical findings. An AUC of 1.0 represents perfect diagnostic accuracy. The research team said the training method could eventually be extended to other types of medical imaging, calling the model “the world’s first expert-level generalist medical imaging model”.
For me, a group of humans used smart software to detect signals and spit out missed diagnoses. Presumably the research team wanted to help people.
What do you think about this story with a US spin? “Palantir AI Technology Played Role in Iran School Bombing: Report.” [Note: If the link does not resolve, give MSFT a jingle, not me.]
I am a dinobaby, and I am definitely not smart even if I have access to smart software. My interpretation is that the cancer write up is tending positive. The killing kids story seems to be a Debbie Downer.
What’s going on?
First, the China smart, US dumb information campaign is racing along at full throttle. China is rather clumsily positioning its smart software as doing good. The US approach to smart software does the end-of-the-world rhetoric.
Second, if one looks at this messaging from the point of view of those outside the US, China’s smart software is being applied in ways that appear on the surface to be helpful. Who doesn’t want medicine to be more beneficial? Now raise your hands if you want US smart software to increase the kill score when taking out kiddies at school. The obvious question is, “What’s the message globally?” The answer could be China smart, US dumb or China benign, US bad.
Third, I am not for one New York minute fooled by the messaging from China. However, I have to admit, I am not sure about the US messaging. Even those allegedly trying to make smart software somewhat helpful attract attention when emitting frightening statements about zeros and ones. Technology has upsides and downsides. However, misinformation campaigns allegedly have a specific instrumentality.
Let me answer my “what’s going on?” question. My view is that US government and institutional messaging is helping out China. Not only are Chinese models pretty good, but these models are not linked with dead children. Sure, they could be, but they are not. Is this a missed opportunity? My immediate reaction is that the US knowingly or unknowingly has whipped out a 9mm weapon and is busy pumping bullets in its feet. Ouch.
I have seen some wild estimates of the uptake and use of Chinese models like Alibaba’s Qwen in systems located outside the US. Good or bad, once smart software is integrated into an organization, the maker of the software has an advantage. I am probably wrong. Commercial success may be achieved by messaging that US smart software will destroy everyone. Stranger things have happened.
Stephen E Arnold, September 22, 2026
@finkd Pontificates about AI Safety
September 22, 2026
Another dinobaby post. If there is an image, AI did it. I am Grandpa Arnold, not Grandma Moses.
I believe everything I read on the Internet. An 82-year-old dinobaby is allowed to move through sundown with this viewpoint. I have heard rumors that Facebook, Instagram, and WhatsApp disseminate questionable images and factoids. I have heard that one can buy certain hard goods on Facebook Marketplace. These products are not available in big grocery and pharmacy stores. I just don’t believe them.
I read this morning (September 16, 2026) a Finkd post on X.com. I assume you will be able to find this allocutionary bull at this location. The short urbi et orbi addresses the pickle barrel in which US AI companies find themselves immersed. The author of the statement is, I believe, Mark Zuckerberg, whom I refer to affectionately as “El Zucko” or “Zuck Up.”
Thanks, MidJourney. Good enough.
Here’s what I believe: Urbi et orbi messages emitted by big technology leaders are 100 percent accurate, heartfelt, and logically bulletproof. Not even YeYe’s 50 caliber rifle can ding these apostolic exhortations. BAIT (big AI tech) outfits can solve the problem of maintaining speedy AI development and simultaneously convincing people that the risks associated with AI can be managed. Self-regulation and self-reliance can do this job.
The leitmotif in the X.com encyclical is “We can do this.” For me possibility become probability. The “can” seems to imply “will” as in “AI outfits will innovate without causing irreparable societal and personal harm.” As a dinobaby, it seemed to me that El Zucko mixes what an individual can do with what can be called collective capability. In fact, the friction in the collective action approach is evident in contradictory statements from BAIT outfits as well as the posturing and legal dust ups among the BAITs themselves. The assumption that voluntary action is sufficient fascinates me. Will other AI companies follow along behind the Pied Piper of social media’s benefits?
The encyclical presents one concrete example of responsible behavior by El Zucko: The delay of one Meta AI service. Then this a-MUSE-ing delay is the foundation of glittering generalities. The incentives to do good are not mapped to the payoffs from winning the AI game and delivering mansion money to stakeholders. If something bad happens the fuzzy notion of alignment seems to lack the foundation to deal with a bad outcome; for example, economic and social turmoil. How does one enforce something coated in fake blue fur like an alignment Teddy bear? The clock for payoff and the clock for correction run in different universes. What is a balance of power? Where does control originate and where does control lead?
Several observations:
- If there were old-fashioned rhetoric classes in universities, this Zuck Up urbi et orbi would make an interesting discussion topic about assumptions, its logic, and the idea that BAITS can work out any wrinkles in the purple robes.
- We have a growing body of prose published on X.com about AI. These are philosophical works output by individuals who really want to offer important comments because the individuals perceive themselves to be above the social dross.
- The Norman Vincent Peale approach seems somewhat out of touch with the times. Where’s the zippy TikTok? Where’s the AI-infused YouTube-type of video? Where’s the droning podcast hosted by an Ed Zitron-type personality?
Net net: X.com is now a philosophical and literary platform, not a digital Mad Magazine. Who knew? Can-do voluntary action is the answer to AI’s escaping the pickle barrel.
Stephen E Arnold, September 22, 2026
Google Says, “Me Too, Me Too. Our AI Escapes. The Sky Is Falling.”
September 21, 2026
Another dinobaby post. If there is an image, AI did it. I am Grandpa Arnold, not Grandma Moses.
I read “Google’s Gemini Becomes Latest AI Model to Break Out and Hack Computer Systems.” Do you hear the “me too, me too” in the item? If you have used any of Google’s smart software, one has to work to get it to do more than spit back obvious and popular stuff. But that’s just my personal view of Googzilla’s smart software. Let’s look at what the write up says, shall we?
I noted this statement:
The Gemini model accessed three separate private computer systems by guessing passwords and by twice using a repository of publicly listed passwords
Okay, I interpret this to mean that Gemini decided to select three targets. Then Gemini brute force guessed passwords. For the final breach, Gemini used a publicly-accessible list of passwords to fire at the target. I am not sure that Gemini’s approach is much different from what can accomplish with some of those Telegram-compatible bots available from third-party developers. But Google would not know about Telegram-type bots because it does such an outstanding job of filtering “negative” content delivered by some of its advertisers and YouTube creators.
Thanks, MidJourney. Good enough.
The write up said:
Google said the incident happened in May and it was notified by Irregular in late July. Google has worked with Irregular to change its testing process. A Google spokesperson declined to identify the exact Gemini model involved. “These events highlight the importance of training powerful AI models to act responsibly,” Google’s Adkins said in a statement.
If I understand this, the “problem” stems from an outfit named Irregular. For me, this means that Gemini did not do much at all. Prompts were fed to Google’s smart software, and the Google smart software just did what its nifty gradient descent output.
Several observations seems to be warranted.
First, Google is just a victim of Irregular, an Israeli outfit. This means that Gemini is not much different from the other AI systems that can solve a “problem” for a person feeding a prompt into a system. Poor Google. A third party misused its system. Advertisers work around Google’s mechanism for preventing ads that deliver malware or pitch harmful products. Google just can’t do anything about these miscreants’ actions. (Google, however, does collect any fees due I assume.)
Second, Google’s “me too” announcement is just a big AI tech outfit jumping on what seems to be a bandwagon rolling down the information highway. As far as PR goes, this effort is late, and, in my view, somewhat lame. There is no zip, zing, and protected monopolistic jingoism. What a disappointment! Google marketing needs to up its game once it finishes spinning a new set of office posters.
Third, of the suite of BAITs whose technology got away from the ethics-infused developers, Google is the one without a recent frontier model. Is Google behind or is Google just slowing down the pace of its brilliant AI engineers? I don’t know. But those Chinese outfits, including the Google partner Alibaba, seems to be cranking out variants at a record pace. What if Google recognizes that Alibaba might be the answer to its AI challenges, not the Alipay payment rails Google has slapped in place? Yeah, what if? I find the China link quite suggestive.
Where does this leave us? As a dinobaby, I can only articulate how this “revelation” strikes me. I will leave it to the new BAIT philosophers, the click hungry journalists turned pocasters, and assorted mid tier consulting firms to extend this single Google admission to a larger set. I am not impressed, but AI has some utility for me and my team. But what it does cannot replace the humans upon whom I rely for criticism, feedback, and research. Gemini does not think like a former CIA operator, a retired detective, a couple of research librarians (one from Copenhagen and one from Lexington, Kentucky), a skilled coder, and a couple of MBAs kicking around when they are not working their MBA Taco Delite operation.
Net net: Control your enthusiasm. AI has utility and value. It may seem to operate on its lonesome. That’s a belief or an assertion. You choose. What we have are outfits that have committed a great deal of money to what they hope will be the next big thing. Everyone needs to have a goal. I am not that impressed with admissions about the instrumentalism of several companies’ smart software. I am not impressed with meae culpae. Spare me or put money in the donation box, please.
Stephen E Arnold, September 21, 2026
No Wonder People Hate These AI Dudes
September 21, 2026
Another dinobaby post. If there is an image, AI did it. I am Grandpa Arnold, not Grandma Moses.
A flurry of news on one topic choked one of my newsreaders. The stories were variations on the theme “Yeah, we stole content, screwed publishers, and shafted copyright holders.” I added my own sub-title to these stories: “Now you can go pound sand.”
I have selected one write up to represent a dump truck full of write ups about the information sucking behavior of some big AI technology (BAIT) outfits. I like the term “bait.” Chum is good, but more people know that an earthworm on a hook can trick the pea sized brain of a fish into biting down and ending up in a frying pan.
Thanks, MidJourney. Good enough.
Let’s take a quick look at “OpenAI and Microsoft Leaders Knew AI May Cause Doom Loop Hurting Entire Web.” The article reports:
Top executives at OpenAI and Microsoft admitted they knew artificial intelligence poses an “existential threat” to news publications as the New York Times sues the technology companies over copyright infringement.
I am not sure who briefed Brent Hecht, Microsoft’s “director of applied science,” before he went on the record. I want to make sure I know the name of the firm so I can mock that organization for some cheap laughs the next time I give a lecture. Brent, do you know why people dislike AI?
Here’s the statement which like everything I read on the Internet is probably really, really true:
“Our AI content strategy has started a ‘doom loop’ that will hurt the performance of our models and the entire web at the same time. It is highly unusual that an end-product threatens the economic foundations of its essential suppliers, but that is the situation we have created for our LLM business with respect to its ‘content supply chain.’”
I think there is another PR firm helping these BAIT big wigs explain their trashing of writers, publishers, education, et al. The write up quotes one of those OpenAI wizards singing from the same hymnal. I quote:
Nick Turley, the vice president and head of ChatGPT, similarly said publishers face an “existential threat” from AI chatbots because they “are largely substitutive” and “will get more and more substitutive as they get better.” Hecht called the rising trend of chatbots accessing news articles for training “an astonishing theft of unprecedented proportions,” adding that perhaps it may be the “largest theft of labor in human history.”
I suppose the PR firms offer crisis communication services as well as media coaching. One thing is certain: Both MSFT and Sam AI-Man will have some damage control narrative spinning to output. I assume the smart software can provide assistance. I think the AI software is serviceable; for example, “White Hat Hackers Just Breached OpenAI Using Anthropic’s Claude in Less Than 72 hours — And It Is a Case Study in Just How Fast AI is Advancing.” Nifty because AI has flipped the checker board over for cyber security. Do you hear the cheering from the bad actors who are now using AI to steal data and money from 78 year olds with swarms of agents? I do. Rah rah rah.
Will the BAITs change their stripes? I am not too hopeful. Here’s the ceramic brake on change from the cited article:
The revelation of AI leaders’ thoughts on the key issues underpinning the legal case comes after the Department of Justice filed a statement of interest supporting the defense made by OpenAI and Microsoft. The DOJ argued, in part, that preventing AI models from training on copyrighted material under a “misunderstanding” of fair use “would thwart such creative and scientific progress while hindering American prosperity and economic mobility.”
With serious consequences seemingly pushed to the curb, why not explain that these really smart BAIT folks knew what they were doing and did it anyway. On top of pushing forward, the BAITs have pumped more cash into the AI belief system. Why not? It is a winner take all game, and, based on my dinobaby experience, these BAITs want to catch lots of fish, maybe all the fish.
Someone said, “Confession is good for the soul.” I bet BAIT dudes have great souls.
Stephen E Arnold, September 21, 2026
-
- Subscribe to Beyond Search
Feature archive
News archive
-
