Looking at AI Layoffs from a Different Angle: Pig Butchering May Benefit
May 27, 2026
![green-dino_thumb_thumb[3]](https://www.arnoldit.com/wordpress/wp-content/uploads/2026/05/green-dino_thumb_thumb3-3.gif "green-dino_thumb_thumb[3]")
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I am not sure “AI layoffs” are the only cause of some financial headwinds in AI Land. However, I did find some of the information in “AI Layoffs Were Supposed to Boost Stocks. They’re Doing the Opposite.” interesting.
The unintended consequences of “good enough” smart software. For serious commercial, medical, and technical applications, AI might just good enough. But to bad actors it is an accelerant. The layoffs do nothing more than provide a pool of angry, capable professionals. Some might be tempted to give online crime a try. Remote work, good pay, no mouse tracking — what’s not to like. Thanks, MidJourney. Good enough to put put some illustrators into another line of work.
The premise of the article is that firing people is the reason that certain companies share prices are headed south. I noted this statement: “New data from CNBC tracking 23 S&P 500 companies that announced AI-linked layoffs found the move is backfiring more often than not.”
The somewhat disjointed write up adds:
A separate Gartner survey of 350 large-company executives found companies cutting for AI reasons weren’t generating better returns than those that didn’t. As it turns out, the highest-ROI companies were found to be using AI for “people amplification” (making workers more productive), not replacing them outright.
The article continues:
The “AI layoff” has become a corporate ritual at this point: cut staff, say AI made you do it, hope investors cheer. But the data says it isn’t working on either end. Markets aren’t rewarding it, and 49,135 workers have lost jobs to AI attribution so far in 2026 alone, nearly as many as all of last year. Even Sam Altman called it out. In February, he said there’s “some AI washing where people are blaming AI for layoffs that they would otherwise” do anyway. So basically, the companies claiming benefits from AI layoffs are pretty much just in trouble and looking to spin it as a positive (“look at how productive we are now!”).
For me, the most interesting RIF story is the one about Meta’s approach to freeing up cash for the firm’s limping Llama LLM technology. Yahoo Finance told me in one of its mash up news stories “Meta Layoffs 2026: 8,000 Jobs Cut in AI Restructuring:
Meta began notifying roughly 8,000 employees Wednesday that they are being laid off, the first wave of a major restructuring the company has framed as necessary to fund its push into artificial intelligence. Singapore-based workers were the first to learn their fate, with layoff emails arriving at 4 a.m. local time, according to Bloomberg. Employees in the U.K. and the U.S. were slated to be informed as their own mornings began. Meta asked North American employees to work from home Wednesday.
My view of the stock connection and Meta’s gutting of people, not just jobs is a consequence of America’s technology oligopolists in general and the BAIT (big AI tech) companies. The supporting players are in this too, but I am just setting them aside in this short blog post.
What’s going on is that AI gained traction slowly and then caught the attention of Microsoft. That company tried to pivot to make AI its go-to tech play. What it did was trigger a PR and marketing war. Companies piled in, flipping the same crazy alarm that Google did. The fear of being left behind as future earnings were sucked down the pockets of the first AI players has created a very crazy environment.
People are being terminated, but it is not clear why. AI projects are everywhere, but obvious winners are tough to find and verify. Professionals are routinely embarrassed by hallucinating outputs copied and pasted without close inspection by lawyers, students, medical professionals, consultants, and more. The past, current, and future cost of inefficient AI compute, algorithms, training, legal fees, and infrastructure are making it easy for glib critics to zap the AI soothsayers. Even small towns are jolted from doom scrolling on the mobiles to protest big data centers that will have some impact on the people living near them.
Now companies are scrambling. A few BAIT outfits are doing the Meta Zuck chest cracking; for example, Microsoft. Others are trying every conceivable trick to generate revenue and “win” the title of top oligopolist in AI Land. Others are just promising breakthroughs like self directed training in the real world. Yep, any day now.
The disruptions are now visible. The explanations are remarkable demonstrations of creative excuse making. What’s significant is that different types of disruptions are popping up. Smart medical outputs are incorrect and can kill patients. Smart agents are useful, not to some businesses. But to bad actors the sketchy agent technology is a like giving a capable bad actor super powers.
I would argue that the layoffs are going to free up some talent to give cyber crime a whirl. For those already engaged in illegal activities, the “good enough” large language model approach to finding vulnerabilities and automating pig butchering is a true gift. Who else sees AI as a perfectly satisfactory turbo charger for online fraud with capable remote workers looking for a gig?
That’s the knock on from AI craziness that does not get attention.
Stephen E Arnold, May 27, 2026
South Korea: Is Total Surveillance Inevitable?
May 26, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I read “Police Plan to Require Cases Opened, Investigated Same Day for All Stalking Reports” about a somewhat interesting policy. Note that the article is about a law enforcement approach in South Korea, not the United States. Before looking at the write up, I want to define stalking as I understand it. Stalking means the behavior of a person who watches, follows, or monitors another individual. The behavior is not a one off; the stalker evidences a pattern of intentional activity. The person stalked realizes something is amiss and then behaves in an out-of-character way; fear of leaving work or an apartment. I want to point out that stalking using social media is a go-to service for some bad actors. Monitoring posts, planting an Apple-type tracking device, or just hanging out in the hopes of spotting the target through a window provide a stalker with tools that Jack the Ripper could not have imagined without the help of AI.
Thanks, MidJourney. Good enough.
So what’s the policy shift?
The write up makes clear:
The [South Korean national] police plan to require cases to be opened and investigated the same day for all stalking reports…
This is a going to be difficult. Stalking, particularly online stalking, is one of the cyber crimes that is in high-growth mode. GenZ criminals may not realize that their doom scrolling and Instagram habits can freak out people if the stalker’s actions are detected. That means there will be a steady increase in the number of stalking cases reported to the police in South Korea.
Second, the idea that “all” stalking cases have to be investigated the day a case is opened. The killer is the “all”. The number of police, regardless of country, is finite. At this time, the law enforcement officers without whom I interact have made it clear that their case loads are increasing. One US cyber crime investigator told me that a job with a policeware vendor was in his future. “I’m burned out,” he said.
Third, as if the “all” was not enough and probably impossible as most “real news” categorical affirmatives are, a law enforcement officer has to initiate an investigation the same day it is “opened.” That’s great if there are officers with slack time. How often does that happen?
I understand the policy. I have no problem with implementing it as long as the investigating officer has time the day the case is opened.
But I want to take this policy in a different direction.
In my opinion, “all” and “same day” will not be what happens. In some cases, the gap will lead to a victim’s suffering serious harm. If crimes continues to rise (particularly the online social media variety), the policy will be wonderful on paper, but in real life, not so great.
The logical solution, in my opinion, is total surveillance. Here’s why:
- A Palantir-style “seeing eye” allows online access to an information lake. That lake can be explored electronically; thus, this type of policeware makes the South Korean policy possible
- Since stalking is one of the online crimes that is in growth mode, the use of total surveillance makes sense. Why roll a vehicle with two or more people to look for a shadow figure? Why use existing cyber investigative tools when more sophisticated systems provide a better way to spot a signal? Therefore, money will be allocated to amp up existing policeware service procurement.
- Only a total surveillance solution can cope with the same day requirement. Any other type of approach introduces delay and hence a policy failure. That means the head of the police will be subject to administrative review. This is not a positive step in a senior officer’s career in my experience.
Net net: South Korea’s stalking policy may set the country on a path that leads to “total surveillance.” I may be wrong. I hope I am. But the risk seems high that administrative machinery will roll directly to this type of solution. Crime is a big problem, and it is getting bigger boosted by — what else? — smart software. Bad actors use bots / agents to do the heavy lifting. Law enforcement is working to catch up. That is why the “same day” phrase is a key signal.
Stephen E Arnold, May 26, 2026
Cyber Security with AI: Do Not Overlook Insider Threats
May 21, 2026
![green-dino_thumb_thumb[3]](https://www.arnoldit.com/wordpress/wp-content/uploads/2026/05/green-dino_thumb_thumb3-1.gif "green-dino_thumb_thumb[3]")
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I read reports about the wonderful work of smart software. It finds bugs. It finds simple mistakes. It finds all sorts of things. Cyber security professionals are working to tap the remarkable power of “Attention is all you need.”
I do want to remind anyone involved in cyber security that interesting new threats arrive daily. One example was Telegram’s announcement that it has used smart software to make programming Telegram smart contracts easier. Let’s assume that this statement is true. Who will be among the first to poke, prod, and explore this lowering of the learning curve for adding intelligence to Telegram-enabled bots, mini apps, and dApps (distributed applications)?
It will not be the frontier model folks. When content turns up and the frontier bros do some training, then the magical Mythos-type cyber security apps will get a boost.
Which employee is the insider threat? Thanks, Midjourney. Good enough.
However, let’s slow the hype train to make sure the wheels are not wobbling.
“Twin Brothers Wipe 96 Gov’t Databases Minutes after Being Fired” does not report about smart software preventing a breach. The article includes a reference to the use of OpenAI’s ChatGPT for guidance. The cyber threat was background checks, questionable personnel processes, and humans with a history of creating digital excitement.
The government agency losing some data was the Department of Homeland Security. But a total of 96 government databases were eliminated after the twin brothers were terminated. The company responsible for the hiring of these two clever lads is identified in the source article. I suggest you locate that firm in case you are working with the outfit. I won’t name the company.
My focus is on the massive craziness about how wonderful LLMs are at finding bugs. That over-the-top response to a function most of the LLMs my team and I have tested can do this type of work. Sure, one has to fiddle around with prompts and being clever with query sequences, but the capability is there. It seems new, unique, novel, and monetizable. Absolutely.
But the big time cyber threats are often the ones created by:
- Slipshod vetting of employees who will have access to digital systems. One cannot do a Google search for an entity like “Janet Louise Lovell” and conclude, “Oh, she’s a low profile person. She’s not a problem.” This means that hiring processes have to shift from amateur hour mode to professional grade investigative probes. Insiders are a big time cyber security risk.
- Management like those lovable MBAs, lawyers, and engineers who run companies doing important work have to manage. I know this is a radical concept, but I think that most senior executives chase sales, pitch for money or contracts, and show up at conferences to watch people the way a wolf tracks a lone sheep. The procedures followed for the termination of the brothers did not cut off system access for both brothers. One slipped through the cracks and deleted the data. That’s a management problem.
- Cyber security providers need to make sure that “insider threat” appears in their marketing materials, is articulated in face-to-face meetings, and included in any explanation of the risks bad actors pose. To compromise an employee, a skilled operator can use specific techniques to gain access. LLM powered cyber security might assist in preventing a problem, but humans are vulnerable. Some need money. Some harbor a grudge. Most don’t want a loved one harmed.
Net net: Senior managers need to manage procedures, processes, and policies. Most don’t want to do this work. What’s the editorial policy at Google? How does OpenAI screen employees? How does Microsoft prevent disgruntled employees from revealing proprietary information? Mythos-type systems are useful, but ignoring insider threats is not a good idea.
Just ask someone at DHS about that disappeared “dhsproddb.”
Stephen E Arnold, May 21, 2026
Pavel Durov: TON Foundation, Where Have You Gone?
May 13, 2026
This is another Telegram-related post.
We posted another Telegram write up the other day. “Durov and the Telegram Foundation” asks a basic hot oven question, “Is the Goat cooked?” Last week, Pavel Durov asserted his greatest of all time business and technical powers. With the toss of a horn, he snagged TONcoin, blockchain plumbing, and product control from the TON Foundation. The Goat cooked essay explores possible consequences in Switzerland, the country in which the TON Foundation was registered. The purpose of the Foundation was to assume responsibility for the Telegram crypto TONcoin. This token was originally named GRAM, but charges of improper fund raising created a legal problem for Telegram and the GOAT. The Foundation was a play to separate Telegram, the messaging service, from the crypto. With the re-taking of TONcoin and related technology and responsibilities, Swiss regulators may have cause to investigate the GOAT, the Foundation, and those involved in what was until recently the marketing and promotional arm of Telegram.
To get the complete essay and some speculation about the legal issues that can arise from this Durov decision making, Telegram Notes’ write up is a click away. Please, be aware, that since active blocking of our Telegram related content, this incarnation of our recycling of some of the material in our new monograph for law enforcement, cyber attorneys, and cyber fraud investigators renders slowly. Be patient. We are doing what we can to provide information that some folks don’t want publicly available.
Stephen E Arnold, May 13, 2026
Telegram News: A GOAT Rolls Over in Iraq
May 9, 2026
Another brief write up about Telegram.
One of my team members sent me this news item today (May 9, 2026): “Telegram Services Officially Resume in Iraq Following Regulatory Pledges.” When I think about countries hassling Pavel Durov (the GOAT of Russian innovation), Telegram (the platform with many crypto capabilities and services), and the shattered remnants of the TON Foundation — I think about France, Russia, and Switzerland. Iraq? Not so much.
According to the write up:
The Iraqi Communications and Media Commission (CMC) announced on Saturday, May 9, 2026, the official resumption of the Telegram application services across Iraq. This decision follows a period of suspension and comes after Telegram’s management provided formal pledges to comply with Iraq’s regulatory requirements and legal frameworks.
A middle eastern government professional is petting a docile, well-behaved, and rule-following pet GOAT. Some goats make great pets. Did you know that pygmy goats have to be watched because if one falls over, it cannot get up and will die in the goat pen because it cannot stand up without assistance? Thanks, MidJourney. Good enough.
It appears that Telegram, the company that for more than a decade resisted giving in to government demands that would violate Pavel Durov’s commitment to free speech and other Silicon Valley-type ideas has changed its tune. The write up points out:
…the restoration of [the Telegram] service is contingent upon the platform’s commitment to Legislative Order No. 65 of 2004. These requirements specifically address content moderation, user protection, and adherence to national security protocols. The commission emphasized that it will continue to monitor the platform to ensure these commitments are implemented effectively in real-time. … all social media platforms operating within Iraqi borders are strictly required to comply with local legislation without exception.
I don’t know much about Iraq, its government, or its regulatory power. On the surface, it seems to me that Telegram is going to provide certain information to the Iraqi government. If this is true, what does this cooperation mean for the alleged bad actors using Telegram to assist Iran in evading Western sanctions. Will these Telegram customers switch to old-fashioned banks and the systems which bird dog possible anti-money laundering and know your customer guide lines?
The news story asserts:
As Telegram becomes functional again, users are expected to see stricter adherence to legal standards regarding shared content, reflecting the government’s broader strategy to balance digital freedom with national safety.
Did Pavel Durov make this decision? Telegram is a one-man or one-GOAT operation. Was the decision made by one of his core team. What are the implications of this decision? The idea of cooperating with a government is, according to Pavel Durov, against his core principles.
In the last week, Durov took back control of the TONcoin and its technical plumbing from the TON Foundation. He also announced that the service charge or “gas fee” for crypto transactions would be reduced to make the Telegram platform the most economical crypto service layer available. Plus, he stated that the Telegram platform would validate transactions faster than other platforms’ service layer.
But Iraq?
I think that Pavel Durov, the GOAT, is using his playbook for defiant actions. If he okayed the cooperation with Iraqi officials, he must see that decision as significant as boosting the perceived value of TONcoin, preparing for the possible Swiss investigation into the dissolution of the TON Foundation, and responding to the stepped up efforts of the Kremlin to shift Russian Telegram users to the MAX messaging system. Plus, there is the upcoming trial in France on a number of alleged serious online crimes.
Maybe the Iraq decision designed to signal that he and Telegram are ready to continue the increased responsiveness to legal requests from law enforcement? Before his arrest in France in August 2024, Telegram was not the swiftest outfit in social media to respond to legal warrants. After his arrest, his willingness to cooperate increased. In fact, two high profile takedowns of criminal activity on Telegram are rumored to have been related to his new interest in cooperation. Many would cooperate, it seems to me, if the alternatives were massive fines and up to 20 years in one of France’s most salubrious prisons. Fleury is a treat I have heard.
In my Telegram Notes post, I want to consider some of the pressures upon the GOAT as a result of his financial decisions, the money in Russia he cannot access, and the hoped-for initial public offering. I think these financial pressures are likely to make Durov’s GOAT-powered decision making somewhat unpredictable.
For now, the Iraq roll over is just one more signal that the GOAT says one thing and does another.
Stephen E Arnold, May 9, 2026
Telegram and Durov: Some Fancy Dancing Kazachok Style
May 6, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Online posts and cyber publications have hinted that changes have been afoot at Telegram. I want to comment about one story from Decrypt.co, an online information service. “TON Surges 36% as Telegram ‘Replaces’ TON Foundation” reported:
Toncoin has jumped on Pavel Durov’s pledge to slash fees to near-zero, with TON meme coins piling on gains of up to 150%.
For some, no service charges or what the crypto professional calls “gas fees” is welcome news. When crypto moves through a series of services for conversion from one coin to another, then to a specialist shop, and finally to a “nominee” or actual owner — the service charges add up. Telegram wants to reduce the pain of TONcoin-denominated transactions.
Pavel Durov demands change. Thanks, Midjourney. Good enough.
The article pointed out:
Illia Otychenko, Lead Analyst at CEX.IO, told Decrypt the rally fits a familiar pattern. “Toncoin has a history of sharp rallies whenever Telegram increases its involvement or signals deeper integration, and this move fits that pattern,” he said. “Right now, it looks more like an early-stage narrative-driven spike rather than a full fundamental repricing.” Otychenko flagged $2 as the key resistance level, a ceiling TON failed to break earlier this year, and warned that momentum indicators are flashing caution.
Got that?
Pavel Durov and his Telegram entities are taking immediate and direct action to address a number of challenges his sprawling empire faces. One of the most problematic is that “value” of the TONcoin. The TONcoin has lost considerable “value” in the last year.
A number of interesting questions arise from the announcement that the TON Foundation is now in flux or limbo. One’s point of view determines how to describe this marketing organization.
Let’s list a few of the questions that analysts and regulators are likely to ask:
- Is the TON Foundation what it said it would be when registered in Switzerland, ostensibly to take over the TONcoin, the TON blockchain, and the cheerleading for the Telegram platform?
- Will the two Telegram-linked NASDAQ companies (TON Strategy Co. and Alpha Compute) strengthen their financial positions or will losses continue to accrue?
- Will Russia’s pressure on Telegram morph into the organization’s finding itself slapped with the label “terror related”?
- Will France exert additional pressure on Telegram to shut down certain illegal activities conducted on the Telegram platform by bad actors?
- Will staff working on Telegram-related projects remain committed to the organization? Will developers building Telegram mini apps and distributed applications defect to other platforms?
- Will US super app platforms attract attention from bad actors who are looking to deploy AI-enabled agents to conduct certain illegal activities?
Net net: Telegram is pushing its “renovation and change” button. In fact, it is not pushing. It is just mashing the button repeatedly at this time.
Stephen E Arnold, May 6, 2026
A Modern Medicine Mixture: Careless Doctors and Half-Baked AI
April 22, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold. I find it interesting that AI detectors identify my writing style as AI output. I suppose I should be flattered, but I just don’t care.
I find two subjects deeply disturbing. One is health care. As a dinobaby, I have an opportunity to interact with doctors and hospitals. I did not know I was an old car in need of intense “care.” The other is smart software. I think that some of the AI types and a few of the United Healthcare-type outfits see AI as an answer to their prayers for more money and more leisure time.
Let’s consider human medical professionals. I think that most of these individuals mostly wanted to do medicine to “help” people. However, stories like this concern me: “Florida Surgeon Indicted after Removing Liver Instead of Spleen.” The guts (sorry, I could not help myself) of the story are:
Grand jury brings manslaughter charge over fatal 2024 operation where patient died on table
Thanks, Venice.ai. Good enough.
Okay, a human doc and an unliving jock. Stuff happens. Imagine how much happier the insurance providers would be if medical procedures were handled by smart software. Keep that in mind, please.
I noted this article in the quirky orange Web site: “AI Chatbots Misdiagnose in Over 80% of Early Medical Cases, Study Finds.” The weasel word for AI lovers is “early.” Until a medical problem “manifests” or “presents” itself, I agree that a skilled doctor has to make an educated guess. Furthermore, the Financial Times is reporting on the findings of a single study. That study and its data may not be reproducible or statistically valid. Nevertheless, the article reports:
The researchers evaluated 21 LLMs, including leading models by OpenAI, Anthropic, Google, xAI and Deepseek. It found that failure rates exceeded 80 per cent for all models when they needed to do so-called differential diagnosis — when full patient information was lacking. The failure rates fell to less than 40 per cent for final diagnoses with more complete data, with the best performers exceeding 90 per cent accuracy.
Okay, close enough for horseshoes.
Forbes Magazine published “Deepfake X-Rays Fool Radiologists In New Study—AI Has Turned Medical Fraud Into A Volume Problem.” This article reports:
Every file In a medical claim is now forgeable.
Let’s assume that human docs make errors. Some of this missteps have downsides like death. The smart software outputs wonky information some of the time. Bad actors can use medical AI to commit serious crime.
Now you have some context for my introductory statement that two subjects trouble me. My human medical professionals can make errors that result in chasing down death certificates. The smart software may provide outputs to a simple medical question like “Should I eat more carrots to improve my vision?” And, instead of writing about the Dark Web and Telegram-linked online crime, I can shift to medical fraud. This subject appears to be an opportunity space for bad actors.
Isn’t modern health care a fascinating professional discipline? Flawed humans and software: Quite a mix.
Stephen E Arnold, April 22, 2026
Telegram Rewrites User Messages and Censorship Blooms Like Flowers in the Township
April 8, 2026
Telegram Notes is almost back online. If there is art in a GOAT post, the image emerged from smart software. Grandpa Arnold is not Grandma Moses. The words, however much they appear to be output from a smart system, have been output by a still living dinobaby with some humanoid features.
Our informal blog Telegram Notes is back online. Recent Telegram-related posts appear to have been blocked, scrubbed, or otherwise disappeared from LinkedIn, a couple of LE and intel-related groups, and from a freedom loving, privacy oriented blogging service in lovely South Africa. No surprise to me.
Filtering and Monitoring?
Microsoft and LinkedIn should know what my team and I have been doing. If you don’t understand the methods used to monitor and respond to information not on the agenda of the world’s best developer of email programs, just navigate to Browsergate and work through “LinkedIn Is Illegally Searching Your Computer.” With these alleged data, one wonders, “How did Microsoft/LinkedIn get smart software so very, very wonky?” And the South African operator? No clue. I learned that my content was “flagged” as AI generated. I am not sure whether to be flattered or baffled.
Thanks, Venice.ai. You did not tell me I was creating an abusive image today. Outstanding. Too bad the illustration is just good enough.
Cape Town Cares
The South African censorship move was courtesy of a free spirit in Cape Town, South Africa. (Those townships on the road from the airport to the central city are a delight. Shabeem thinking may be prevalent outside of these residential spaces.) I know that this dinobaby is not the swiftest 82 year old prowling the rest home’s grounds, but it is interesting to think that I am relying on smart software to produce snappy phrases like “write up” and “leadership” in my idiosyncratic way. I even have a GenX or GenY explanation of the content removal. If you want to check out where you too can have your content sanitized, navigate to the cute privacy first just censored BearBlog at https://bearblog.dev/.
Back Online with Telegram Notes… Sort Of
We have managed a new post about Pavel Durov’s Telegram. The write up explains that Telegram has deployed an AI editor in its Messenger container. What’s interesting is that when a Telegram user enters certain phrases, the Chinese large language model changes user content to an acceptable Chinese Communist Party form. You can read that at https://shorturl.at/GhSST.
We haven’t worked out the listing of articles so each appears with the most recent at the top of the stack. There are some definite spacing weirdnesses, and one of my semi-capable team is working on that issue as well.
Observations
A few dinobaby thoughts zipped slowly through what passes for my brain these days:
- When writing about a service like Telegram and posting that information in the US, censorship happens. The vendors insist it doesn’t, and I know that vendors especially like the BAIT (big AI tech) outfits never, ever fabricate or stretch the truth.
- Whatever angle we are taking on Telegram, Pavel Durov, the TON Foundation, TONcoin, TON Strategy Company, AlphaTON Capital, and the cast of characters associated with this sprawling system — someone, somewhere is taking the trouble to block the content.
- Online has become an interesting place. It is populated by Softies, Cape Towners like Herman Martinus, MBAs at LinkedIn, and BAIT guardrail developers.
I still find people like Moti Cristil, Oskar Hartmann, Manny Stotz, Yuri Mitin, and the GOAT himself Pavel Durov fascinating professionals. When you are 82, I hope you find a hobby as fulfilling as mine. Most of the stories about Manny, for example, do not get the timeline for his Kingsway plays straight. Few know about the estimable elite-only Equium Club which operates in places where movers and shakers move and shake. Not too many people have paid attention to the interesting career trajectory of a Ph.D. who worked for Cambridge Analytica and now serves as the president of a company engaged in AI compute leasing and online games named AlphaTON Capital with Enzo Villani, a former NASDAQ professional assisting the firm. The share price of ATON (the Mitin / Villani company and not the Russian bank of that name) was about $0.32 per share on April 7, 2026. That share price is below the NASDAQ threshold for a listed company.
The people in the rest home think I do real work. Nope, I have a hobby, and it is attracting attention from LinkedIn and the fellow who thinks his Bear Blog product is a garden. I think I understand. Gardens have to be pruned. I wonder if Herman’s AI identification system thinks this post you are reading was generated by one of the large language models. And LinkedIn / Microsoft, how can you have so much information about LinkedIn users and you continue to output truly peculiar software. Outlook in space? Nope, a giant technology company lost in space and flailing because its tether had a quality problem.
Oh, well, back to my hobby.
Stephen E Arnold, April 8, 2026
Smart Software Seems to Lack a Capability: Adaptation to That Which It Was Not Trained
April 8, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
William James, the American thinker much loved by first year psychology students, coined the phrase “a certain blindness.” As I recall, the idea is that each humanoid cannot perceive certain things. Scammers use the principle to get money out of grandmas and lonely widowers. Smart people with money instinctively know that they do not have any blindnesses whatsoever.
Smart software operates on the principle, probably because of the limitations of the training sets and the Rube Goldberg machine built from algorithms that power artificial intelligence. I am not sure how one fixes a humanoid who believes he or she has 360 degree sightedness or a smart software system crafted by BAIT outfits. Oh, BAIT is my lingo for “big AI tech.”
I read “AI Can Beat Chess Grandmasters, But It Can’t Adapt to Modern Video Games.” If the write up is spot on, the implication is that when a humanoid does something not in a training set, the smart software is lost in space. The write up says:
AI is still pretty bad at handling a new video game it has never seen before…. According to researchers, many of AI’s biggest gaming successes are based on systems that are finely tuned to one specific game. In those defined boundaries, AI can basically become superhuman. But as soon as there are slight changes to the rules or environments, its impressive performance can collapse.
As Jack Benny used to say, “Yipes.”
The article points out:
The research paper adds that reinforcement learning can produce impressive results, but acceptable goals are only achieved after millions or billions of simulated runs. So the system becomes an expert in the exact situation it is trained for. But all of this falls apart when any changes are introduced. Even something as simple as shifted colors or repositioned objects on a screen can break it.
I can visualize the responses from the BAIT outfits now. The jibber jabber will boil down to denial and misdirection. And why not? Which BAIT outfit wants to have investors and stakeholders shout, “You misled us” or “You are mendacious” or “You are a crook.”
The write up adds:
LLMs (Large Language Models) do not solve this either. NYU [researcher] says they perform surprisingly poorly on unfamiliar games. When it does start doing well, this is usually in custom game-specific scaffolding to interpret game states, manage memory, and execute actions. Strip that extra support away, and performance drops fast.
Interesting. But the AI push is that smart software is the next big thing. If BAIT outfits build it, people will come. Well, that’s the theory. Ignore the surveys that suggest a significant number of people are wary of smart software. What will happen when smart software and smart systems get the sub optimal answer. That will be exciting for some.
Stephen E Arnold April 8, 2026
Social Links: A Fraud Fighter
March 27, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Ivan Shkvarun is the CEO and founder of Social Links. Years ago, I would have described this system as “policeware.” Today many non-law enforcement and intelligence professionals want to integrate disparate data. The firm’s current positioning strikes me as providing open source intelligence with a dashboard and modules. The purpose of the system is to protect an organization from AI threats. That’s a modern positioning. Specialized software firms have to find a way to explain their product in a very short, snappy elevator pitch.
An online publication called Pulse2 interviewed Shkvarun in, “Social Links: Interview With Co-Founder & CEO Ivan Shkvarun About Fighting AI Fraud.” I learned that Mr. Shkvarun is dedicated to building tools that can verify and assess information. The origin story of Social Links was a deepfake video of former President Obama that zipped around the Internet in 2017. Social Links was built to restore people’s trust in data. Social Links’s product line includes AI Defender, a modern digital risk platform.
Shkvarun was asked how his work has global impact:
“ ‘If we look at the rise of software-as-a-service two decades ago, SaaS initially represented only a marginal share of the global software market. The transformation that followed was not driven by explosive growth in overall IT spending, but by a shift in how value was created and delivered. Over roughly fourteen years, SaaS expanded from approximately 2% of the market to more than half, fundamentally reshaping the industry.’
He adds:
‘Artificial intelligence now sits at a similar inflection point. Today, the global software and IT services market is estimated at around $10 trillion, while AI represents only a fraction of a percent of that total. Even under conservative assumptions, AI-driven revenue could exceed $12 trillion within the current decade.’”
Social Links includes functions that help a licensee deal with fraud. Payment fraud can generate substantial revenue for a bad actor. Continuous effort is required to cope with the problem. What makes Social Links different from other companies is its expertise, positioning, and strategic focus. The company continues to grow while maintaining a devoted customer base because it delivers trustworthy and reliable products.
The companies relies on webinars to showcase its interface and the system’s functionality. If you scout around, you can sign up for a webinar.
Whitney Grace, March 27, 2026
-
- Subscribe to Beyond Search
Feature archive
News archive
-
