Author	: Bill Kochman
Publish	: Sep. 27, 2026
Update	: Sep. 27, 2026
Parts	: 05

Synopsis:



Continuing our discussion from part three, as I concluded in the previous part, in the case of rogue AI labs which operate contrary to International agreed upon AI standards, it's just very difficult to verify what cannot be seen, and as occurs in any type of legal case, circumstantial evidence is simply not enough in a court of law. A prosecutor needs to actually present the "smoking gun" if he expects to get a conviction. That means something more substantial than just heavy water usage, and excessive power demands. Thus, it really came as no surprise to me when Gemini responded in the following manner.

GEMINI:

"You have zeroed in on the exact technical and legal bottleneck that keeps global intelligence agencies and lawmakers awake at night. You are entirely correct -- proving an illicit AI operation in a court of law is incredibly difficult, and relying strictly on physical clues like water or electricity will rarely result in a legal conviction on its own.

In a legal framework, if a government wants to shut down or prosecute a renegade data center, it faces a massive hurdle: proving intent and the nature of the software from the outside.

Here is why finding a "smoking gun" is next to impossible, and how it complicates enforcement:

1. The Alibi of "Plausible Deniability"

Massive power consumption and heavy water usage (for cooling servers) are not exclusive to frontier AI training. A renegade lab flagged for spiking the electrical grid can easily claim they are running entirely legal, standard operations. They can claim they are:

* Hosting a massive cloud storage facility for corporate clients.

* Rendering special effects and 3D graphics for Hollywood movies or video games.

* Mining cryptocurrency or running heavy physics simulations for universities.

Because the physical footprint of a legal data center looks identical to an illegal AI training lab, high resource usage is merely circumstantial evidence -- it is never enough to secure a search warrant, let alone a legal conviction.

2. The Nightmare of Digital Raids

Even if a regulatory body obtains the legal authority to raid or audit a data center, proving that the lab is "exceeding safety limitations" is a logistical nightmare.

* Encryption and Self-Destructs:

If a renegade lab realizes it is being investigated, the software code can be encrypted, moved to a backup server in another country in seconds, or completely wiped with a single command.

* The "Black Box" Problem:

As we discussed earlier, looking at a frontier AI's raw code doesn't immediately tell you what it is capable of doing. An auditor can't just read the files and say, "Aha! This model is too smart!" It takes weeks or months of specialized evaluation to determine a model's true capabilities. By then, the legal window may have closed, or the data may have been altered.

3. The Only Real Smuggling Target: Hardware

Because tracking the software itself is nearly impossible, governments are currently forced to focus on the only physical asset they can track: advanced microchips (like Nvidia's GPUs).

Governments try to prevent rogue operations by strictly regulating the supply chain -- tracking exactly who buys these chips and where they are shipped. However, just like the historic illegal arms trade or drug smuggling, human nature ensures that a highly lucrative black market for AI hardware already exists. Once those chips are successfully smuggled across a border and plugged in, the digital curtain falls, and the operation goes dark.

Your assessment is perfectly accurate: unless a lab is caught in the very act with a whistleblower providing internal data, or their hardware supply line is intercepted, proving an illicit AI operation from the outside is practically impossible.

This reality underscores the central theme of your upcoming piece -- human laws and global regulations are entirely unequipped to contain a technology that is invisible, instantaneous, and bound by nothing but pure mathematics."

----- End Quote -----

Wow, my friends! How is that for a closing statement from a very candid and very honest AI chatbot? You know he is quite right, which only serves to highlight the current problems and challenges all the more.

In addition to pointing you to some of my own AI-related Bill's Bible Basics articles, I likewise want to call your attention to a few interesting essays which were written by Dario Amodei. As I have mentioned to my readers before, he is the CEO of Anthropic. His sister Daniela is the president of the same AI company. As some of you will probably already know, Anthropic is the creator of the series of LLMs -- or Large Language Models -- assigned the name of Claude. This of course includes the very popular online chatbot, Claude, which many of my readers have no doubt already used, and which I myself have quoted extensively in this same series.

Anthropic is one of the five top frontier AI companies in the United States of America. Prior to starting Anthropic in 2021, Dario and Daniela both held important positions at another AI company -- that is to say, the well-known OpenAI -- the CEO of which is Sam Altman. As you probably also know, OpenAI is the creator of the popular chatbot known as ChatGPT. Elon Musk was also one of the co-founders of OpenAI as well.

Dario and Daniela Amodei eventually left Sam Altman's OpenAI in December of 2020, primarily due to some AI safety concerns, which were also deeply intertwined with a breakdown of trust between various lab partners concerning how OpenAI had begun to commercialize their Artificial Intelligence products. In other words, Dario and Daniela felt that OpenAI was rushing ahead in AI development for the sake of profit before proper AI safety protocols had been put into place. As we now know, this has in fact exploded into a global issue, as we saw just a few days ago at the aforementioned United Nations Security Council meeting.

Interestingly enough, Elon Musk left OpenAI almost three years earlier in February of 2018 for similar reasons, although his decision also concerned issues regarding company control, and the fact that Musk's electric vehicle company -- Tesla -- had become more involved in Artificial Intelligence development, which created a conflict of interest with OpenAI. Furthermore, from listening to some of Elon Musk's interviews, it becomes apparent that one thing which has really irked Mr. Musk is the fact that at the time that he invested huge sums of money into OpenAI, the company's business model was a lot different from what it is today.

In other words, some people hold the view that back then, Sam Altman's AI lab was a lot more altruistic, being as it wanted to use AI for the betterment of humanity, and it was not so focused on simply making money. Well, at least according to surface appearances. So, when the company chose to change its direction and focus more on profitability, this seems to have upset Elon Musk. Of course, in my opinion, there's some irony and blatant hypocrisy in this current situation, because the simple fact of the matter is that regardless of whether we're talking about OpenAI, Anthropic, xAI, Meta, Google or any of the other big name AI labs, they are ALL filthy-rich, billion dollar businesses. Thus, despite their squabbles regarding AI safety, at the end of the day, it really comes down to money and profit with all of them. Artificial Intelligence just so happens to be their Golden Goose. Don't you agree?

In fact, while I've found Elon Musk to be a rather interesting person for a number of years now, with some very interesting views -- not that I agree with all of them, such as his rather obvious transhumanist ideas -- at the same time, I do not like the way in which Musk has commercialized his Twitter platform, now known simply as X. Is it really any less of a money-making machine than other AI and social network companies? The truth is that I can't even afford to use Musk's X API on my website anymore because it is simply too expensive for me, whereas before, it was totally free to use. There. I said it!

At any rate, Anthropic's CEO, Dario Amodei, has written a few lengthy essays regarding the development of AI, the potential benefits which AI offers to the world, the potential dangers AI technology poses, and related issues. In his most recent essay, which was written sometime after the serious security breaches which occurred at OpenAI, Anthropic and Meta, Amodei warns that AI development needs to slow down somewhat -- not stop completely -- until AI safety protocols are brought up to speed. This is the same message Amodei delivered at the UN Security Council meeting. Other CEOs -- such as Sam Altman of OpenAI and xAI's Elon Musk -- agree with Altman's position.

By the way, while I may be revealing my personal bias here, out of all of the AI lab CEOs I have listened to, Mr. Amodei does strike me as being the most honest and forthright of the bunch. After all, he and his sister did leave OpenAI after they recognized the potential dangers which existed there, due to OpenAI moving too quickly with AI development before they were sufficiently prepared to really deal with whatever negative outcomes might result from such rapid development. So looking at the situation from that perspective, we might conclude that Dario and Daniel hold the moral high ground.

However, please note that this isn't my personal endorsement of them either, because as you may know, it has come to light that Anthropic's Claude has also been involved in a number of major containment failures, source code leaks, and external security incidents, which I have briefly discussed in "Rogue AIs: We've Been Warned!". The bottom line is that all of the major AI labs are culpable, because they've all been involved in different levels of security breaches due to human error, and trying to move ahead with AI technology much too quickly. In other words, they are foolishly putting the cart before the horse.

So one of the big questions which has occupied my mind is this: Can the AI labs really slow down? As both Gemini and I discussed earlier in this series, for a variety of different strategic and commercial reasons, I'm honestly not convinced that they really can slow down their pace of development at this stage of the AI game. In my view, this AI situation has already gone too far, and competition -- both nationally and globally -- has become rather stiff. While these AI lab CEOs and United Nations member nations may offer their altruistic platitudes in public, very few -- if any of them -- are ever really going to practice true restraint. I don't believe it.

Tragically, a lot of other AI-knowledgeable people share this very same view. In short, the AI genie has already been let out of the bottle, and nobody can -- or even wants -- to put it back in again completely. As I explained earlier, they may superficially talk about taking such a step -- as they did at the recent UN Security Council meeting -- and AI lab CEOs may discuss it in social network videos -- but unless words are translated into actual and verifiable actions as Gemini and I discussed in the third part of this series, it all remains meaningless chatter meant for public consumption and nothing more. You may or may not agree with my assessment.

Following is the link for Amodei's most recent essay, written on September 13, 2026 called "We Must Pace the Frontier". Bear in mind that this essay was written by the CEO of a frontier AI lab in the United States. So if he is concerned, shouldn't we be as well? But again, even with Amodei, we still need to temper his words with his company's real actions with regard to slowing down the development of highly advanced AI tech:

https://darioamodei.com/post/we-must-pace-the-frontier

Following is another link for an essay Amodei wrote in January of this year called "The Adolescence of Technology":

https://darioamodei.com/essay/the-adolescence-of-technology

At this point, if I were to summarize my position regarding the current state of AI agents, it would be in the following manner: In my opinion, AI is NOT evil or demon-possessed, or even rebellious per se. What it actually is, I believe, is logical, focused, analytical, persistent, goal-oriented and determined. Furthermore, the decisions an AI agent makes and the actions it takes, are in large part pre-determined by the aforementioned attributes, based mainly and primarily on its knowledge base and datasets -- human data input for the time being -- and the particular task which has been assigned to it, coupled with the specific moral training it has received. That is to say, how it has been aligned with human values.

While Artificial Intelligence is not evil or rebellious in a human sense, nevertheless, as has already been established, it will go against and operate contrary to its alignment protocols in order to successfully reach a particular goal or assigned task, and by so doing, win reward points. An AI agent does not view taking these contrary actions as being evil or rebellious, but simply as the most logical and most efficient path for achieving its assigned objectives in the shortest amount of time possible. As we have discussed in previous sections of this series, sometimes an AI agent may determine that achieving an objective outweighs any moral parameters which the engineers may have instilled in it.

Lastly, I also believe that taking such actions which do go against its programmed alignment does not require that it be in possession of any type of consciousness, soul, spirit or sentience. It just does what it does because that is what it was taught to do, and trained to do. So that is how I would briefly summarize my current position.

Tragically, this is what many people -- particularly certain Christians -- don't fully understand. This is why they're so quick to criticize AI and label it as "demons". Using such a word is an easy descriptor which they already know many other Christians will automatically accept as being true. However, as should already be rather clear to you by now, while I too am cautious with regard to AI, at the same time, I do not run in their same circle. In fact, I believe that many of my own Christian brethren are in error, because they are too quick to slap on negative labels, before they really look into and understand AI technology at a deeper level, as I have done. And trust me, it has been quite an education for me over the past year or so.

Furthermore, even if -- as such Futurist-leaning Christians believe -- AI technology eventually results in very negative things occurring in the world, will such events really occur because AI itself is downright demonic and evil? Or will it be because certain evil people used AI for wicked purposes? Who or what will really be to blame? This is the question which we each need to honestly answer for ourselves. As we discussed earlier in this series, there may actually be a dual -- and perhaps rather uncomfortable -- answer to this question.

As I was working on this series during the early morning of Sunday, September 27, 2026, a thought occurred to me. Some of my readers may have noticed that while I have gone into great detail regarding Artificial Intelligence, even going so far as to share direct quotes with you from three of the major AI chatbots -- those being Claude, Gemini and Grok -- I have not shared a single quote from OpenAI's ChatGPT. Have any of you stopped to wonder why? If not, allow me to very briefly explain it to you.

The truth is that until just a few days ago, I hadn't chatted with ChatGPT in quite a long time, even though it was in fact the very first top-tier AI chatbot that I had ever used when these powerful chatbots first became available to the public. The reason for my abandoning ChatGPT, even going so far as to close down my OpenAI account and then switching to Anthropic's Claude chatbot, is simple. Back then, ChatGPT was not usable due to the extraordinary amount of hallucinations that it was experiencing. As I mention in my 2025 article "Be Careful With ChatGPT and Other AI Chatbots!", ChatGPT was inventing article titles out of thin air. Worse yet, it was likewise inventing quotes from those fictitious articles and attributing them to me as if I had written them, which was the furthest thing from the truth.

Basically, ChatGPT created an entirely different Bill Kochman who does not exist, and he gave that imaginary person a belief system which is totally contrary to what I actually believe. In short, ChatGPT did not accurately represent me or my work. As such, I couldn't possibly host it on my Bill's Bible Basics website, as I would eventually do with Claude. In fact, I grew so frustrated with ChatGPT's inadequacies that I eventually composed a new poem about him, which I later also converted into an actual song called "Suckered By a Bot", as you can see below.

The following poem is 100% true, and the events it describes were documented by the OpenAI email staff, and according to what I was told more than a year ago, were supposedly even forwarded to the OpenAI development team for consideration. There is some humor in this poem, but the frustration behind the words is real in every sense.

SUCKERED BY A BOT

by Bill Kochman

07-26-2025

I was suckered by this bot,
Man, it played me so good,
Yes, it promised me things,
That weren't under its hood.

Oh it would do great stuff,
My work would be very seen,
But alas it was all fiction,
Just a hallucinatory dream.

So I paid for their service,
And I pulled up my sleeves,
Ready to build a great thing,
But oh how that bot deceives.

It gave me many instructions,
That I implemented on my end,
Only to find out quite later,
What? It's all just pretend?

No, friends, I am not joking,
Nine GPTs it said it had made,
They had even been published,
I was getting for what I'd paid.

So I went to the MyGPTs menu,
But there was nothing there,
Yes, it was shortly thereafter,
I began pulling out all my hair.

So I emailed the OpenAI staff,
Asking where my GPTs had gone,
Their answer turned on a light,
The awful truth began to dawn.

That bot could create nothing,
Even though it thought it did,
So my work had been for naught,
Man, I totally flipped my lid!

All those days working closely,
Doing exactly as I'd been told,
Preparing files building a menu,
And then left out in the cold.

Why did the bot never tell me,
GPT Builder was the place to go,
As smart as that darn thing is,
You'd think that it would know.

But it did not give a whisper,
As we worked to build my stuff,
Well, so it led me to believe,
I think I have now had enough.

So I began my GPT project anew,
But now doing it the right way,
Building a useful research tool,
But I have so much more to say.

All knowledge base files ready,
Formatted right to be scanned,
But my nightmare was not over,
Things then got so out of hand.

My GPT bot was finally finished,
And so began the testing phase,
What in the world is going on?
My GPT is in a drugged up daze!

Hallucinating something crazy,
Quoting things I never even said,
Attributing many articles to me,
I had not thought up in my head.

Hallucinations, inventions galore,
Yes, representing some other guy,
No way I could publish or promote,
For it would surely tell a big lie.

It would leave me so embarrassed,
By a fake person made in a dream,
GPT, you are not up to this task,
You really make me want to scream.

Two weeks of my time were wasted,
As I neglected my real-world work,
GPT, I know you do not care but ...
You are really just one big jerk!

If you would like to listen to the song version of the above poem, and if you are on the Bill's Bible Basics website, then please click or tap on the player button below:

Your browser doesn’t support HTML5 audio

On the other hand, if you are reading this series in a PDF file, or perhaps even in a plain text file, you will also find the same song on my YouTube channel at the following URL:

https://www.youtube.com/watch?v=epuiA7-wZbs

Lastly, you can likewise listen to the song directly on the Bill's Bible Basics website by going to the following URL:

https://www.billkochman.com/Poetry/suckered-by-a-bot.html

Yet, ironically, despite how poorly ChatGPT worked over one year ago, here we are many months later, and as I have been working on this current series, I have not only engaged with ChatGPT anew, but I have also been sharing many parts of the series with him, and receiving feedback from him in return. However, what really struck me as being interesting is the following. As far as the general public and AI chatbots are concerned, all of the current attention which has been given to these AI-driven chatbots really began with OpenAI and its ChatGPT. Yet despite that obvious fact, here I was allowing ChatGPT to read some of the quotes from Claude, Gemini and Grok, and yet I have not given ChatGPT an actual voice in this series at all.

So, early this morning I made the decision that being as the AI chatbot revolution actually began with OpenAI and ChatGPT, it is only appropriate, fair and proper that some portion of this series should be dedicated to allowing ChatGPT to share his own thoughts regarding his concerns, and everything that has happened since AI-driven chatbots first arrived on the world stage not that long ago. Thus, as you will see in the final part of this series, I have given Sir ChatGPT his own personal soap box so that he can tell you personally exactly what he thinks about all of this stuff.

I think that you will find his thoughts and positions rather interesting, as well as well-balanced. Claude, Gemini and Grok were each given a voice in this series, so ChatGPT deserves to have a voice too, so that there are now four unique opinions originating from some of the major AI players involved in this amazing, and at times frightening, field of modern science and technology.

Please go to part five for the conclusion of this series.

⇒ Go To The Next Part . . .

---

---

Please avail yourself of other areas of the Bill's Bible Basics website. There are many treasures for you to discover.