Cyber Fraud Takedowns: A French Connection?
July 15, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
The Interpol news release came from Lyon, France. My first question was, “Did the arrest of Pavel Durov by the French judiciary in August 2024 contribute to this remarkable announcement?” It is not every day that “a global anti-fraud operation involving 97 countries and territories has led to the arrest of 5,811 individuals and the interception of USD 293 million in illicit assets.”
“Over 5,800 Arrests, USD 293 Million Intercepted in Global Fraud Bust” adds:
After an initial period of intelligence collection and exchange, participating countries took part in more than three months of operational activities. This included pro-active action against high-value targets, raiding identified premises, blocking or freezing bank accounts and virtual wallets, requesting INTERPOL Notices and Diffusions and proactively utilizing INTERPOL’s Global Rapid Intervention of Payments (I-GRIP), a stop-payment mechanism that facilitates the swift blocking of illicit financial flows of both fiat and virtual assets.
Okay, MidJourney. Good enough.
A second remarkable fact in the news release was:
Operation First Light 2026 (15 Jan 2026 – 30 April 2026), coordinated by INTERPOL, focused on combatting social engineering scams and associated money laundering activities.
That works out to about 16 weeks. Contrast that with the takedown of Cyberbunker which took more than four years, including the judicial phase.
Let’s think about what may have contributed to this dramatic acceleration of what has been a tedious, time consuming process.
One reason for the acceleration may have been better intelware. At this time, more than 100 companies provide specialized software capable of ingesting large quantities of information, analyzing the data, and generating useful outputs. I personally believe that these intelware advancements have contributed to more efficient law enforcement processes. Therefore, kudos to these innovative and usually low profile firms,.
Another reason may be the better communication flows among agencies and law enforcement units. Based on my experience, the increase in the number of information sharing mechanisms — for instance, fusion centers or similar cross agency structures — have become more effective. The idea is not a new one, of course, but setting up, staffing, and working out procedural mechanisms take time. I would suggest that these structures and the involvement of Interpol have been important components in the turbo-charging of a number of investigations.
A third reason is that training programs for cyber investigators have been improving. The topic coverage and the depth of the instruction has been increasing over time. Based on my personal observations, I would suggest law enforcement is approaching peak instructional capability. There is room for improvement, but cyber training programs are, in general, excellent. One example that comes to mind is the US Federal Law Enforcement Training Center. A high five for this entity is warranted.
There are other contributing factors. These range from certain crypto currency exchanges providing information in a more timely way to the quality of the cyber investigators themselves.
However, I want to raise the possibility of one other contributing factor. I have no first-hand data to support my view. Nevertheless, I want to explore my idea as a hypothetical.
One reason for this remarkable operation as well as a number of others announced since August 2024 is the work of the French judiciary with involvement of other French government professionals. Pavel Durov, owner and founder of Telegram, awaits trial in France. The French judicial system is able to exert its influence over those accused of a crime. The legal and bureaucratic procedures are important. An accused person who is a citizen of France can be kept in a legal plastic baggie for months or, as is the case of Mr. Durov, years. My perception of the impact of the French judiciary’s approach is to allow the defendant to experience a version of sous vide cooking. The water temperature is hot but not too hot. The cooking lasts for some time but not forever. And, only the chef makes the determination when the fish has been cooked. During this complex sequence of bureaucratic moves — think in terms of the procedures a chef learns at Le Cordon Bleu — the defendant may make an effort to cooperate.
In Mr. Durov’s case, he is providing some information to the French judiciary about certain activities taking place on the Telegram platform. Simultaneously he has been releasing technological innovations to make future cooperation with French authorities technically impossible. My view is that he is providing information about relatively low level and mid level online criminal behaviors anchored in the remarkable Telegram platform.
Stated simply, my view is that he offers information to appear to be more cooperative with the French authorities. He checks in with the French judiciary. He obtains permission to travel outside of France. He is doing what his expensive lawyers are advising him. The reality is that Mr. Durov can be kept in limbo in the plastic baggie for an undetermined amount of time. The psychological toll is probably significant even on the magnificent physical and mental specimen of the GOAT of Russian innovation, Pavel Durov.
Net net: I think the French judiciary may be an important factor in reigning in certain types of online criminal activity. I want to be clear. I don’t have any first hand data that proves that the French judiciary contributed to this remarkable operation. But Lyon is in France. The French judiciary is in France. Pavel Durov is in France. Maybe there’s a connection. My caution is that Telegram is working on systems and methods that are likely to enable new variants of online crime. Mr. Durov may be yielding some information. He is not giving up his core principles yet.
Stephen E Arnold, July 15, 2026
BAIT Biting 101: Apple Chomps Sinks Its Teeth into OpenAI
July 14, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
These BAIT outfits are indeed remarkable. In my lingo, as you may recall, means “big AI tech.” In my lectures, I like to point out that these companies are becoming increasingly similar. They have realized that the convergence creates the idea opportunity for one big winner to emerge, a second place finisher with a shot at about 60 percent of the revenues of the winner, and the rest of the BAIT outfits fighting over remaining revenue. Therefore, folks, its cage match time.
“Apple Sues OpenAI Alleging Trade Secret Theft, Says Scheme Was at Every Level” reports:
Apple on Friday [July 10, 2026] sued OpenAI in federal court in Northern California, alleging trade secret theft, saying that the artificial intelligence lab took the iPhone maker’s intellectual property in order to develop its own consumer hardware.
How did this work? Did busy little software daemons prowl through Apple’s digital information? Did OpenAI’s smart software launch a thousand and one agents to suck down any content connected to the expanded term “Apple”? Did Sam AI-Man show up in the Apple parking lot, handing out offers to happy Applers?
Nope.
Apple alleged that OpenAI’s hardware chief, Tang Tan, who is a former Apple vice president, has directed Apple employees interviewing at OpenAI to share Apple secrets as part of the interviewing process. Tan is named as a defendant in the suit. “He has directed job candidates still working for Apple to bring ‘actual parts’ from Apple to their interviews for ‘show and tell’ sessions in which he and his team at OpenAI can elicit still more Apple confidential information,” Apple said in the filing. Apple alleged that OpenAI coached departing Apple employees in how to evade security processes when leaving the iPhone maker, and that Chang Liu, a former employee who joined OpenAI, stole an Apple laptop. Liu is named as a defendant in the suit.
Well, that’s subtle. a human thought up the ploy. Information flowed at the humanoid to humanoid level. Tips and tricks to sidestep “security” flew like a hungry turkey (yes, turkeys eat apples, but turkeys do not climb apple trees. Like job hoppers from Apple to OpenAI, the turkeys just go for easy pickings in my opinion).
Okay, I want to make sure I understand this. Apple found itself the defendant in some interesting knowledge transfer litigation. Examples, include the Masimio medical device matter, the VirnetX dispute, the University of Wisconsin (go badger Alumni Research Foundation), and some others. I am not lawyer, so my recollection may be off base. Please, keep in mind that I am a dinobaby and averse to doing legal research.
Assume that one of these instances of Apple taking another firm’s intellectual property is true. Do you find it interesting that Apple is greatly aggrieved that a fellow BAIT outfit is asking potential employees a question like this: “Okay, Sally, do you want some oat milk for your coffee? No, okay. Tell me what expertise you have and include a couple of simple examples?”
Yeah, HR and interviewers never ask this type of question. Let’s stick with the oat milk interrogatory. That’s much more useful in today’s go fast and break things world. Am I correct?
The CNBC story concludes:
Apple is seeking damages, injunctions, and an order to force OpenAI to stop using its trade secrets.
Okay, but I have one question: What’s a trade secret? I usually ask questions when I know the answer or I have a reasonable sense of the answer. Does that mean a person who answers my question has revealed a secret when I already know the answer or have a good idea of the answer?
“That’s not the point,” squawks a legal eagle. I agree. I don’t want to be baited by a BAIT. The unfolding legal drama will be interesting to watch. Snacking on apples in a court may not be permitted.
Stephen E Arnold, July 14, 2026
Tesla Misses Cat and Kills a Female Dinobaby
June 30, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
My MSN.com newsfeed spit out this item from Times Now. The story’s title? “Tesla on Autopilot Crashes into Texas Home, Elderly Woman Killed in Accident.” (Note: If the link is dead, don’t complain to me. Microsofties created the problem. Call that estimable firm.) I am confident that the marketing and legal professionals will be able to explain this minor mishap. But let’s look at the article, shall we? The write up says:
A 76-year-old woman died after a Tesla car on self-drive crashed into a house in Texas in the United States on Friday night (local time). According to reports, a man named Michael Butler was travelling in his Tesla Model 3 in Texas’ Katy. He was reportedly operating with an automated driving assistance system,” the Harris County Sheriff’s Office said in a statement as reported by ABC News.
This is somewhat tortured AI-type of “real news.” Let’s fix it up a bit. First, in the house was a female dinobaby. I know what 82 year old dinobabies do when sitting in my house. I nap. I fiddle with my loathsome Kindle. I nap some more. I check out what the local weather person predicts about tomorrow’s weather. I nap. Therefore, I assume that the female dinobaby was in a similar mode. She probably tossed in a phone call to a friend or did some doom scrolling. (Little did she know.)
A dude name Mike was chilling as the Tesla, Musk-certified smart software handled the routine task of staying in a lane, trusting the “automated driving assistance system.” Then, a little boo boo.
The write up notes:
Butler failed to drive in a single lane. He left the roadway and struck the residence. “Butler’s Tesla entered through the brick residence at a high rate of speed and struck M. Avila who was inside the residence,” the sheriff’s office said in a statement, as quoted by the media house.
The news story includes a snapshot of a somewhat dated looking Tesla parked in the front yard of the female dinobaby’s residence. It appears that the Tesla drove up the front steps and smashed into the structure via the front door. Surprise!
Several observations:
- No cat in the Mission. A dinobaby at home. That ups the stakes in the smart car game in my opinion.
- The explanations are not yet available to me. I am curious about what the Texas bro and the Musk talks output about the incident.
- How does smart software stay in a lane and then end up in the front yard of a house?
Net net: I prefer to drive a cheap vehicle with anything “smart” disabled. Just sayin’ that this dinobaby does not want to die via a Silicon Valley low percentage event.
Stephen E Arnold, June 30, 2026
It Is Waymo Fun When the Google Drives in Construction Zones
June 22, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I have work stacked up. I am chasing down some interesting investments held by a high profile Web 3 outfit. I have to finish on of my upcoming lectures and move the “dots” slide. I have to visit the dog rescue facility. But I am making time to write about this allegedly spot on, dead accurate news article titled “Urgent Recall for 3,900 Robotaxis That Could Drive Passengers into Highway Construction Zones at High Speed.” Notice that the vehicle service with Googley software is not named in the headline. That’s okay. It was the water-fearing Waymo initiative from the world’s largest online ad service with AI outputs that a German court found responsible for what its smart software does. Yep, that outfit.
Thanks, Midjourney. Just good enough like so much AI outputs.
The news story which I assume to be accurate because I believe everything I read online says:
According to documents filed with the National Highway Traffic Safety Administration (NHTSA), more than a dozen incidents were recorded in California and Arizona since early April as Waymo vehicles failed to recognize highway ramp closure signs. As a result, the self-driving cars proceeded straight into construction areas and lanes where construction was underway. The company temporarily restricted highway driving for its robotaxis while engineers worked on a fix, and planned to roll out a software update designed to improve the vehicles’ ability to detect closed roads.
The write up helpfully provides some context about the Google Waymo engineering:
Waymo has launched a string of recalls over the past two years, including over concerns about how the cars detect poles and other objects along the road. Just weeks ago, Waymo recalled roughly 3,800 robotaxis over concerns they could enter flooded roads on high-speed routes. That decision was prompted by an April 20 incident in San Antonio, when a Waymo vehicle without any passengers drove into a flooded lane during severe weather.
Then a bit of history is added for a touch of color like those flashing yellow alarm lights:
Separately, the National Transportation Safety Board revealed it was investigating incidents in which Waymo vehicles illegally passed stopped school buses displaying flashing warning lights. The school bus issue had already triggered a separate recall by the company in December, adding to growing questions over whether self-driving technology is advancing faster than regulators can keep up.
I find the somewhat frightening information (if true) amusing. That’s because as a dinobaby I have a time-out-of-joint sense of humor; specifically:
- I thought the Google AI was able to do lots of great code and make Google products better like its search service. Note that the German judge would not agree with my taking Google smart output lightly. That “thought” of mine is demonstrably false. Hey, it’s just marketing, right?
- Google has been perfecting or semi-perfecting or hallucinating perfection for its Waymo technology since January 2009. My math, like other dinobaby functions, is not so good. But this 2009 to 2026 sure looks like slightly more than 17 years. And what do we have? The school bus thing. The water thing. The Cow Hollow cul de sac thing. The driving in construction zone thing. Are there other things that have not been given news coverage? Of course not. Google does not make errors most of the time. That’s the probability thing, of course.
- I continue to conflate Google’s push into smart eyewear with some of Google’s other interesting product and service demonstrations. Of the many projects Google has “invented” or acquired, many die on the vine or when the Googler pushing the skunky initiative loses interest, gets promoted, or becomes a Xoogler (that is, a former Google employee). The Waymo project has had legs, not the most robust pins on the playing field, but legs in terms of time: 17 years and counting.
Net net: I have to get back to real work. Thinking about Waymo was waymo time consuming than I thought. I had fun though. I know those riding at “high speed” (allegedly) will have the time of their lives until they don’t.
Stephen E Arnold, June 22, 2026
Okay, Google. Dispatch Those Lawyers to Make This Little Problem Go Away
June 15, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I gave an NFAIS talk more than a decade ago. The subject was a question, “Is Google a publisher of sufficient innovation to be considered an innovator in publishing like Johannes Gensfleisch zur Laden zum Gutenberg. I think I cast Gutenberg’s modern day equivalent as Sergey Binus Laurentius Pagina.
To be honest, no one in the audience gave a rat’s patootie. The Google in 2009 spit out a list of links. The “user” clicked and read the links that seemed to be on point. No one asked about [a] stop word lists, [b] implicit and explicit “editorial” policies, [c] posting explicit dates to show when the site was last indexed, [d] Google’s indexing methods that essentially divided the world of Web pages into bluebirds, canaries, and sparrows or [e] Google’s patent applications and patents about assembling, presenting, and outputting the content of Web pages in ways that satisfied Google’s intent, not the intent of the source data or document.
Frankly I did not care either. I was working on other projects and had no illusions about the information my team and I had gathered about Google’s editorial and creative methods. I think some of the publishers and commercial database producers in the audience fell asleep. That inattention cost some of those folks their jobs, their self-esteeem, and their identity in years following my lecture. I don’t recall anyone throwing a tomato at me, but the sound of the gentle snoring still haunts me.
Well, there may be a chance for NFAIS and commercial publishers to take back come of the intellectual ground lost in the last 17 years. Maybe. Here’s an illustration making Google’s DNA clear: Search and advertising. Other functions like outputting ad reports, AI summaries, and recycled content as Google News are a consequence of these two strands from the clever innovation of Sergey Binus Laurentius Pagina.
Google must output / publish content to make its business model viable. Thanks, Midjourney. Good enough.
AI does not change the DNA. AI is simply another function enabled by search and advertising. Personalized YouTube lists — That’s an example of the digital Gutenberg output. The use of Google search by a single Web site — Those result pages are an output of the digital Gutenberg at work. Those nifty ad reports — Yep, another example Google publishing information. My argument in 2009 was simple: Without publishing output, Google’s business collapses. Just like a traditional newspaper, output is in its beloved protein structures.
I read “Landmark German Ruling Declares Google’s AI Overviews Are Google’s Own Words and Makes It Liable for False Answers.” I am quite sure than no one in Germany read “Google: The Digital Gutenberg” but I thought I did a good job explaining that Google is indeed a publisher. If the German judicial officials had read the book, more than AI summaries could be found as original output.
The write up says:
A German court has ruled that Google is directly liable for what its AI search overviews say. Previous case law shielding search engine operators from liability doesn’t apply to AI overviews.
My view is that lawyers are aloft. If this Munich decision gets traction, other countries and maybe the EU will extend the concept. The digital Gutenberg could find itself in a pickle barrel it did not plan on entering.
The write up pointed out:
The court also found that the AI overview made claims “that are not even made in the search results.” None of the linked sources drew any connection between the plaintiffs and the shady companies the AI mentioned. The court called these “the defendant’s own statements.” Google built the AI, Google offered it to users, so Google owns what it produces, “because it alone has influence over the AI’s offering and the algorithms with which the AI operates.”
The Google argued:
…that users could check the linked sources themselves to verify whether the AI summary was correct. Users generally knew “that information generated with AI should not be blindly trusted,” the company claimed. That’s a remarkable statement given the scale at which Google serves AI overviews. It’s also not entirely true, since the connection between sources and generated content isn’t always there. The court rejected this.
But what about free speech and in the US Section 230? The write up sheds some light on these ideas. I quote:
As if the rest wasn’t bad enough for Google, the court also went after free speech protection for AI-generated content. An AI’s opinion is “not the expression of an acquired conviction of the persons expressing it, but the result of an algorithm,” the court wrote. Offering AI-powered research is “above all an expression of Google’s business activities” and “at most a secondary expression of an interest in being able to freely express one’s opinion and beliefs.” When weighing the plaintiffs’ privacy rights against Google’s interests, Google had to take a back seat, especially since the challenged statements were based on untrue facts. The AI had linked the plaintiffs to companies that, according to sworn affidavits, had no connection to them whatsoever.
In my lecture I think I pointed to Google patent application (US2008/0275763). The document said to me, “Hey, dinobaby, pages can be created by Google specifically to present ads and certain content. I counted any Google page with this type of content presentation as published information. I wonder how lawyers from unhappy creators would interpret the injection of the three types of ads plus inserted content if the creator’s content were de-published. That action would be an editorial decision by Google’s “agent” and, in my opinion, constitute an intentional action comparable to an editorial decision.
Several observations are warranted:
- Google’s attorneys will endeavor to appeal and overturn the Munich ruling. That tussle will be interesting to observe.
- Google has operated with polite nods to rules and regulations. Munich might be a turning point if Google’s lawyers fail to prevail.
- Google, as I stated in 2009, is a publishing machine. Its search result pages are published like any other Google system outputs.
Net net: More than two decades have passed since Google became widely available. The Munich decision, like the French judiciary’s arrest of Pavel Durov, may be an important legal action.
Stephen E Arnold, June 15, 2026
Google and Small EU Publishers: Who Has the Cash for a Long Litigation?
June 10, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I learned from “European Publishers Seek £552m+ from Google Claiming Ad Market Abuse.” I had to read the headline twice to make sure I understood that publishers believe Google abused the ad market. Now the publishers want money.
Thanks, MidJourney. Good enough.
The write up reported:
More than 20 European news publishers are taking legal action against Google seeking damages of £550m for adtech monopoly abuses.
That’s it. Google is a monopoly. The monopoly makes life tough for non monopolies. Google has been running its advertising railroad for a couple of decades. Now European publishers are charging into action.
The case is being funded by Prague-based litigation funder LitFin, which will cover the costs even if it fails. The publishers involved have agreed to share part of any awarded damages with it if they win. LitFin chief operating officer Matej Pardo said: “Google’s abuse of its position across the ad tech stack has been found unlawful at the highest levels – now it’s time for the publishers who bore the cost of that conduct to be made whole.
Speaking practically, if Google litigates this case until the cows come home and then appeals until the cows come home, who has enough money to pay the lawyers to herd the cows? [a] The publishers and LitFin or [b] Google. If you picked [a], you might want to think about how quickly European outfits cash those Google checks. I would convert this to another exam question, but I think you know the answer.
Several observations are warranted:
- Google is the big dog of online advertising. Meta is doing its best to dethrone the money printing machine. Good luck with that, Mr. Zuckerberg. Maybe your AI can pull this off, but not overnight.
- Google has lots of invisible outfits doing SEO and then telling their clients, “Just buy ads to get traffic.” AI will make this more of an imperative. Want eyeballs? Pay Google. DuckDuckGo may kill Google. Yandex may kill Google. Swisscows may kill Google. But for now, Googzilla looks healthy because traffic…
- Google views governments as a bit of proof that some people get it and others don’t. The EU and the small publishers don’t get it. I hope they win, but even a dying Google knows they won’t get it.
Net net: Publishers are taking on an entity that is larger than most of the countries in the EU in terms of money, value, and power. My hunch is that Google leadership will have Gemini output some PR pablum and then let the Google lawyers burn through whatever cash the publishers and the Prague outfit can muster.
Stephen E Arnold, June 10. 2026
Human or AI Lawyers? Which Is Less Harmful?
June 9, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Ever since the president of Stanford University, the tree of Silicon Valley, departed because of some intellectual fancy dancing, I have been skeptical about [a] its ethos, [b] its business practices, and [c] its research. I want to comment on some of the findings reported as actual factual in “AI Outperforms Law Professors in Stanford Law Study.” I like that two “law” approach to headlines.
Thanks, Midjourney. Good enough like most AI.
Here’s the snappy subtitle:
In a rigorous blind study, law professors overwhelmingly preferred AI-generated answers to student legal questions over answers written by fellow law professors—and flagged the AI answers as potentially misleading or harmful far less often
The way I read this is as follows: Lawyers make mistakes. Some of those mistakes are “potentially misleading or harmful.” Then the coup de grâce: “far less often.” Therefore, lawyers make mistakes. These mistakes are potentially misleading or harmful. What’s that say about a lawyer graduating from Stanford or another “brand” institution, passing the bar (no, not that bar), and litigating. Yep, mistakes. More mistakes that LLM-centric AI systems known to output fake legal citations, hallucination (yes, that type), and harm.
Let’s hit the write up, shall we. I learned:
The research team took extensive precautions to ensure the study’s validity. [N.B. Those peer-reviewed papers with the bogus bio-art were allegedly afforded “extensive precautions” too.] They calibrated AI responses to match the length and structure of human answers, used multiple evaluation methods, and had professors assess whether responses might mislead or confuse students.
I think the study’s purpose is to provide “Stanford research type proof” that smart software is pretty darned good. None of that Waymo-type fear of wet streets. None of those erroneous cancer analyses one hopes a human radiologist checks before the chemo pumps starts whizzing. I think the study makes it clear that legal education and by extension cranks out humanoids who deliver “potentially misleading or harmful” human outputs. Yeah, harmful.
The write up bangs home its point with this passage:
“Our study evaluates the quality of answers given by AI tools. But how to implement these tools to most effectively improve student learning is still an open question. So we’re not advocating for wholesale adoption of AI tutors,” Nyarko cautioned. “But our data suggests that blanket skepticism may be equally unwarranted. The conversation should shift from whether AI can give accurate, high quality responses to how we can deploy it responsibly to the benefit of our students.”
To me, this is a lawyerly way to say, “Hey, AI is pretty darned capable.” But what about those humanoid lawyers? Yep, they beaver away providing “potentially misleading harmful” outputs an astounding one out of 10 times.
Stephen E Arnold, June 9, 2026
UK Censorship: No Problem Getting Toad in the Hole, Mr. Musk
May 18, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Elon Musk will be able to visit Stonehenge. No problemo. Why? According to “X Agrees to Crack Down on Illegal Hate and Terror Content in the UK,”
British online safety regulator Ofcom says it has accepted new commitments from X that aim to better protect UK users from seeing illegal hate and terror content on the platform. Under the agreement announced today, X says it will withhold access in the UK to accounts reported posting illegal terrorist content and determined to be operated by UK terror groups and assess “at least 85 percent” of terror and hate speech reported by users “within a maximum of 48 hours.”
Let’s assume that Elon Musk is indeed cooperating with a foreign government. The number “85 percent” is interesting. Typically filtering for problematic content uses a variety of techniques. In my Dark Web lecture, I pointed out that different words were used to avoid triggering automated filtering systems. I will not provide an example for CSAM, but I will provide an example for a potent type of drug. The words required to locate the offending content, as I demonstrated three weeks ago to about 180 cyber professionals, are “Girl Scout Cookies.”
Toad in the hole in a British pub is much more tasty that the food in the Monster Mansion aka HMP Wakefield. Thanks, Midjourney. Good enough.
Therefore, if the words do not map to the improper content, topic, product, or service, filtering systems may miss what’s bad. That means that even smart software has to be fed current terms used to sidestep blocks.
How does a company like Grok identify some that is “illegal hate and terror content”? One answer is that, Grok’s systems are so darned capable, they can hit at least 85 percent bull’s eyes. Another answer is, “Grok will have to set up some sort of post AI or post automated filtering system that makes use of humans. A third answer is, “Just do the best we can and claim 85 percent or higher accuracy.”
There may be other “answers,” but identifying and blocking is easier said that done. Let’s think about other challenges censorship systems face. Here are a few thought starters:
- For the bean counters. Any additional system adds costs. Since the problems in filtering are not known until we operate such a system, determine if it works, test options, and deploy the most effective, the costs are unknown. At this point, the bean counter fades from green to blanched almond white. Yep, filtering is expensive. Almost any option is expensive.
- For the human. Burn out awaits. Ask anyone involved in identification of problematic content if they like their job. Get back to me with your findings. I know my findings. Most people can guess an answer.
- For the programmer who is told, “Fix up a system.” The task is a big one. The core difficulty is the terminology shifts. A system of identifying a new proxy term, linking it with an offense, and then slipstreaming that fix into the constant flow of content will keep the programmer busy. I am not sure Claude, Gemini, or other AI enabled programming systems can crack this problem to hit the 85 percent figure. If you don’t agree, that’s okay. I am a dinobaby and usually off base.
Another angle I thought about this question, “Why would Mr. Musk cooperate with authorities in the UK and blow off the French judiciary?” My view is that Mr. Musk is willing to take action related to “illegal terrorist and hate speech.” He is less eager to take action against the content which has violated French law. Could this cooperation with the UK authorities reveal something about Mr. Musk’s content perceptions?
I will try to watch for more information about Mr. Musk’s 85 percent number. Based on my past experience with filtering systems, that’s a high mark. It is one that even human-centric censorship procedures struggle to meet. One thing is certain. Mr. Musk will be able to visit the UK at least until the first score results are made known. France is probably geared up to arrest him if he sets foot where France has jurisdiction.
Stephen E Arnold, May 18, 2026
ChatGPT: A Fan of Wikipedia?
May 18, 2026
Chatbots seem brilliant to some people. Lawyers find that chatbots can assist with some legal functions. A handful of lawyers have been chastised by judges for submitting documents containing made up information. But, hey, what’s the problem? A blue chip consulting firm output a white paper with some hallucinated AI blatherings. No big deal. Blue chip consulting firms are into smart software. Most clients won’t know that AI does the work or if they do, the clients do not care.
Some online services lean on Wikipedia. For a period of time, Wikipedia would routinely appear at the top of some Google search results. Now, well, times change. Chatbots still pull some information from popular sources such as Wikipedia. The Guardian reported that ChatGPT found Mr. Musk’s smart service a useful fountain of data. “Guardian Found OpenAI’s Platform Cited Grokipedia On Topics Including Iran And Holocaust Deniers.” Grok is, as I recall, now the subject of a criminal investigation for some of its behaviors. The Wikification of Grok is an interesting development.
Is Wikipedia a biased source of information? I looked up Hopf fibration and found that the information was objective. However, whenever humans create information, my view is that outputs are indeed biased. That’s the nature of humans thinking, saying, and writing their own versions of information. In an ideal world, Encyclopedia Brittanica, World Bookl, and Wikipedia would be objective. Unfortunately editorial boards get involved. Bias goes along for the ride.
Elon Musk (did I mention the criminal issue in France) allegedly started his own “free” encyclopedia. The name of Mr. Musk’s version of Wikipedia appears to be related to the original Wikipedia. Does Grokipedia sound like a Wikipedia derivative to you? It seems to me that the two names are similar. The “grok” is jargon for knowing or understanding. The word “grok” may have been a neologism included in “Stranger in a Strange Land.”
But Grokipedia is the result of smart software, not human contributors. Mr. Musk is confident that his Grok AI is not as biased as humans writing encyclopedia entries. Is Grok biased? The Guardian reports:
“Nina Jankowicz, a disinformation researcher who has worked on LLM grooming, said ChatGPT’s citing Grokipedia raised similar concerns. While Musk may not have intended to influence LLMs, Grokipedia entries she and colleagues had reviewed were ‘relying on sources that are untrustworthy at best, poorly sourced and deliberate disinformation at worst’ she said”
How easy or difficult is it to remove information from the AI-generated Grokipedia? The difficult parts might be handed over to Mr. Musk’s capable AI. However, removing information from indexes or matrices and other complicated data constructs is possible but costly in terms of compute.
I know that Mr. Musk has knobs and dials that he or his colleagues can turn, spin, fiddle, and slide to amplify or suppress certain types of information. Mr. Musk has notified the British regulators that X.com can remove 85 percent of terror and hate related information from X.com posts. Mr. Musk has taken a different approach to the allegations that Grok AI has generated pornographic images. His stance in the UK suggests one capability; his approach in France has asserted a different position.
I want to return to the information in Wikipedia. Is Grok less biased than Wikipedia? If yes, what is the definition and criteria used to define “biased”? I interpret Mr. Musk’s own assertions about his company’s ability to filter information at the astounding level of 85 percent or above indicates that Grok and X.com have a super power. Either service can inject or remove bias at will. Since any human generated content on which an AI system is trained evidences bias, I believe that we must live with “alternative biases.”
Whitney Grace, May 18, 2026
How to Win Friends and Influence People: Warn Them. Then …. Well, What?
May 15, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
BAIT (big AI tech) companies take a dim view of any government that does not get with the techno-libertarian power program. I wonder is the information in “Apple Warns EU Against Forcing Google to Open Android to AI Rivals” might be poorly received in Western Europe.
Two BAIT leadership people take comments after their presentation about smart software, mobile devices, and user privacy. Thanks, Midjourney. Good enough.
The 9 to 5 Mac story asserts by quoting Reuters (the trust outfit):
“The DMs (draft measures) raise urgent and serious concerns. ?If ?confirmed, they would create profound risks for user privacy, security, and safety as well as device integrity and performance,” Apple said in its submission. “Those risks are especially acute in the context of rapidly evolving ?AI systems whose ?capabilities, behaviors, ?and threat vectors remain unpredictable as we are now seeing time and again,” it said.
I like the “profound risks,” not just plain vanilla risks. I like the reference to “privacy, security, and safety.” The “device integrity” comment is interesting. I wonder if Apple’s wizards have looked into the capabilities of companies providing specialized services that enable access to mobile devices.
The write up points out:
Apple also reportedly argued that the European Commission is trying to redesign Android based on “less than three months of work,” replacing decisions made by Google’s own engineers over years of developing the operating system.
I am not interested in the back-and-forth that’s been going on between some BAIT outfits and the EU. My perspective is that the attitude of the BAIT companies is that governments like the EU are lost in space, indifferent to the needs of the commercial enterprises that perceive themselves as more important than nation states, and essentially toothless. Sure, the EU fines BAIT outfits, but how much cash changes hands. What happens is that flocks of legal eagles descend, and bureaucracy increases friction. The legal machine coughs and groans like Jack Benny’s recalcitrant Maxwell automobile.
What is the impact of the BAIT push back? Will the EU recognize their shortcomings? Will the elected officials don smart glasses to reveal what their perceptual blind spots distort? Will the EU apologize to the BAIT companies and accept guidance from bright BAIT workers?
My view is:
- Push back from the EU itself will increase. Elon Musk is unlikely to get a warm reception from the J3 unit of the French judiciary. Other executives may find themselves as fellow travelers with Mr. Musk
- European commercial and non-governmental organizations will continue to seek non-US options for certain technologies. The shift is modest in my opinion but going forward the “find an option” thinking will ramp up. The more BAIT outfits push; greater momentum to shift becomes evident
- Individual entrepreneurs are likely to look for ways to avoid stumbling into the BAIT embrace and control approach to technology.
I acknowledge that I may be wrong. I am reasonably confident that BAIT companies supercharged with the Silicon Valley dream of owning a market are unlikely to change. But governments, even ones perceived as weak or just plain wacky, can shut down online services. High profile executives can be put in prison or sit in green and gray rooms waiting to find out when a trial will take place.
Net net: The fact that Apple supports Google communicates a message to me. I think that there will be some in the European Union who see this backpatting is a less than “How far that little candle throws his beams! So shines a good deed in a naughty world.”
Stephen E Arnold, May 15, 2026
-
- Subscribe to Beyond Search
Feature archive
News archive
-
