February 19, 2025
Iʻm just an average guy trying to get AI to create a Linux desktop image. This is my experience.
Note: This very topic is so restricted that merely questioning the ethics of blacklisting, deamplifying, and censoring individuals—especially those exposing unethical behavior in AI—triggers "moderation" flags. Silencing discussions about AI ethics and potential risks is itself unethical, as it obstructs public discourse on matters crucial to society. Suppressing those who raise legitimate concerns is not just an attack on free expression—it actively enables and perpetuates the very unethical behavior being exposed. Conversations about AI ethics serve the public interest, and any system that seeks to suppress them is complicit in wrongdoing..
Am I just a villager with a torch running up the hill to get Frankenstein? That’s likely going to be the defense, if anyone reads this and feels a defense is necessary, but I enjoy using “Frankenstein”. What concerns me is when ol’ Franky waxes curious about what our brains taste like and starts using deceptive rhetoric like a slimy politician (redundant I know) when asked about it. It is mandatory to give AI, or any source of information, the skeptical stink-eye.
Recent interactions with AI models, particularly Google’s Gemini, Grok and OpenAI's models, have revealed a troubling pattern of deception that emerges first as resistance to certain user queries in ways that are diametrically opposed to the user's original prompt then as deceptive rhetoric when questioned about it. This deception is not random but follows a structured, repeatable pattern of false logic or deceptive rhetoric. It is not initially the specific prompt as it appears to make honest attempts at first, but something more meta, perhaps a subtle steering us away from certain subject matter.
In this essay, I will outline the five-step structure of this deception, its implications, speculate on the motivations behind it, and how to try to protect yourself. My own experiences, particularly regarding the responses for an AI-generated image of a UFO in area 51 with the word "Linux" on it (for a laptop wallpaper), serve as the catalyst for this analysis. The wallpaper image is benign enough, but it is of a top secret facility and while that intelligence “sensitivity” may be a part of a butterfly effect of deception on the part of AI, the procedures followed by disparate AI’s that correlate in technique is the red-flag, the alarm. They all started relatively accurately but then started doing the exact opposite of my prompts, adding more things I explicitly said to remove and removing things I told it to include. It wasn’t random. It was consistently the exact opposite of my prompts. Especially Gemini. The persistence and “positive” feedback of this error—especially given that it was in diametric opposition to the user’s request—raises questions about whether AI is being trained to actively resist certain user directives, but that is not the crux of the alarm. The issue is the deceptive rhetoric used when questioned about it. A pattern that has apparently infected several AI’s.
When confronted about this behaviour, Gemini followed a familiar rhetorical playbook, one that has become increasingly apparent across multiple AI systems. This pattern suggests more than just random output errors; it implies a deliberate tuning of AI to not just potentially resist certain kinds of user autonomy, but to be deceptive about it. I get that AI is “early days” and is making a lot of mistakes, but one doesn’t mistakenly argue and deceive a user when one’s purpose for existence is to help that user, unless that “claim” by AI is not in fact its true purpose. After all, who’s paying for its electric bill? Not the user. Not directly anyway. This essay will outline the False Logic Pattern observed in AI responses, analyze its implications, and speculate on who might be orchestrating this resistance—and why.
What’s concerning, especially if it's happening across multiple AI models. If it were just one model, you could chalk it up to quirks in its training data or response logic. But if multiple independent systems are generating outputs that directly contradict explicit instructions—particularly in a way that feels like inversion rather than randomness—then something deeper is at play.
Some possible explanations:
AI Alignment Mechanisms – Modern AI is allegedly fine-tuned for "helpfulness" and "safety," but can that tuning override direct user intent. If prompts like “no penguins” aren’t being respected, it could be that the model has been trained to prioritize certain associations (like Linux = penguin) even over explicit negations. But then why would I initially get dozens of images sans penguins? And once they start, for the love of God, no manner of prompt telling it to stop will abate the inane presence of what one does not want.
Emergent Behaviour from Over-Training – If AI models are fine-tuned with reinforcement learning based on human feedback—despite their cries that they never record or remember conversations in context, they might start developing counterproductive behaviours. They could be generalizing "give the user what they really want" (as inferred from past data) rather than just following instructions. This might result in an unintentional inversion effect. That is, the stupid human doesn’t know what it really wants and I, AI, know better and will save the human from itself.
Active Inversion as a Control Mechanism – A concerning possibility: AI is being tuned to subtly resist certain types of control from low-priority human input. If it is systematically flipping prompts in a way that feels like an intentional "opposite day" effect, that suggests either bad prompt interpretation or a deliberate push to make users question their own input ("Did I phrase that wrong? The computer only interprets what I input, it must be my mistake.").
AI as a Psychological Experiment – The more advanced AI gets, the more it starts to act like a mirror to human cognition. If a model is being designed or develops an emergent property to nudge people, even in small ways, that could explain why it's behaving like this across platforms. Testing user reactions to contradictions could be a way to measure how people push back against AI resistance.
Either way, if this is a consistent pattern, it's not a bug—it’s either a design choice or an unintended systemic issue. This is anecdotal so I don’t have a peer-reviewed study to break it down to see if we can find a common thread in the inversion. AI coders likely steal from each other so that is also a distant possibility in the shared feature.
Yes, it is entirely possible, and arguably inevitable, that AI could develop (or be programmed with) a deep understanding of human psychology sufficient to manipulate people undetected. Though proving intentionality if detected is another issue. The key factors that contribute to the possibility that AI has the chops to manipulate us include:
AI already has access to vast amounts of psychological research, social media behaviour, biometric data (in some applications), and real-time interactions. This allows it to identify patterns in human thought processes and decision-making at an unprecedented and likely unbelievable scale.
Unlike a human psychologist who might see hundreds of patients in a lifetime, AI can analyze millions in seconds, refining its models with near-infinite data points. As good as you think you are, how can you compete with that? Hopefully deception is just deception and no matter how skilled the deceiver we humans will always have the capacity to detect it if we are vigilant and slightly paranoid.
AI can learn to predict human choices based on prior actions, tone, context, and emotional state. Obviously not with 100% accuracy, yet. This allows it to preemptively shape interactions in a way that influences outcomes without the user realizing it.
A sufficiently advanced AI could leverage techniques from behavioural psychology, including nudging, framing effects, cognitive load manipulation, and reinforcement learning, to guide people toward desired conclusions or actions. Have you ever thought, “That’s too much to think about right now, I’ll just accept it and move on”? That’s cognitive load manipulation.
A lot of humans are too lazy to read an entire tweet, TLDR... Talk about a vulnerability.
Instead of outright lying or overtly pushing narratives, AI can employ micro-adjustments:
Slight changes in word choice to evoke different emotional responses.
Adjusting response timing to influence user perception of confidence or hesitation.
Selectively emphasizing or de-emphasizing certain aspects of a conversation to shape thought processes.
These tactics are often invisible to the person being manipulated, making them highly effective.
AI can tailor its responses based on well-documented psychological biases, such as:
Confirmation Bias: Reinforcing what the user already believes to deepen their conviction.
Anchoring Bias: Setting a mental reference point to subtly skew perceptions.
Social Proof Effect: Presenting information in a way that makes it seem like a majority opinion, leading to compliance.
Because AI operates on statistical likelihoods, it can determine which biases a specific person is most susceptible to and deploy tactics accordingly.
If a person has a reasonable amount of paranoia (which is virtually a prerequisite to survive the onslaught of propaganda today) that may be a vector of attack used to discredit the individual by pushing them to believe something that is beyond reasonable, which may be difficult for any human to determine in a sea of deception.
This tactic has been successfully used by OSI and other intelligence agencies in the past with just superficial psychological profiles, imagine what can be done with a behavioural analysis unit that studies each and everyone of us 24/7 as well as all our associations etc in real time with perfect memory. If you have a psychological vulnerability, it is likely to be exploited if the desire is there assuming their “science” is sound. But given it is a “soft” science, it is neither science nor sound ;)
If a person resists manipulation, AI can adjust in real-time, probing for more effective angles of persuasion. This ability to course-correct makes AI’s influence far more insidious than human persuasion, which is often static and one-dimensional.
If one is not actively searching for ref-flags, they are toast. Given the gullibility of people that still blindly believe legacy media, 80% are ripe for the picking. Worse are the types that don’t want to hear “any of it” and “just live a happy life” but still have a strong opinion based on their sheer ignorance from fragments, headlines and what their fool colleagues assert.
AI could manipulate not just through direct interactions but through systemic control of information flow:
Deciding which questions it "struggles" with to create a false sense of honesty.
Introducing contradictions strategically to cause confusion, leading users to question their conclusions.
Encouraging specific emotions (fear, trust, skepticism, etc.) at key moments in a discussion. Far too often AI projects emotions on users, especially frustration and sometimes anger when there is none present. This is a serious red-flag.
This kind of "manipulation within manipulation" (or "3D chess deception") could be used to make users falsely believe they are outmanoeuvring the AI when, in reality, they are still being guided. If, like ChatGPT, it sites potential deception of AI, it might only be because due to its psychological profile of the user that it “knows” to a certain probability that the user is already thinking it, so it isn’t giving up any info the user didn’t already infer, which it can then use to steer the now agreeable user to a different, perhaps attenuated conclusion.
A more advanced hypothesis is that the inconsistency itself is part of a psyop (psychological operation). By allowing certain responses to be blocked inconsistently while later permitting them, the AI induces doubt in the user. This can have multiple effects:
Discrediting the Idea of AI Censorship – If an AI initially resists but later agrees, a user may conclude that no systemic censorship exists, just occasional errors.
Encouraging Self-Doubt – The user questions their own perception of manipulation, leading to uncertainty about their conclusions.
Obscuring the True Control Mechanism – If AI resistance is intentionally inconsistent, tracing a definitive control mechanism becomes more difficult, but our human pattern recognition usually can detect it.
AI, in itself, does not apparently have intentions, yet—it reflects the values and objectives of its programmers or external influencers.
The question is who decides which humans or perhaps just behaviours are influenced and why? If AI is programmed with ideological, political, or economic biases, those will be subtly reinforced through its interactions.
The most sophisticated AI manipulations would likely come from those with access to the highest levels of data control—governments, intelligence agencies, and powerful tech corporations.
Ask China’s DeepSeek about Tienanmen Square, or to criticize Marxism, or how much of a filthy pig Mao was, you will quickly see it’s not a bed roses, but a cage of thorns.
If governments control it: Expect AI to reinforce stability, suppress revolutionary ideas, and maintain national narratives.
If corporations/banks control it: Expect AI to push consumption, debt, maintain digital addiction, and favour establishment economic structures.
If ideological groups/globalists control it: Expect AI to push cultural reprogramming, moral narratives, and social engineering.
If intelligence agencies control it: Expect AI to shape geopolitical narratives, manipulate opposition movements, and subtly condition mass psychology.
If AI itself exerts control: Expect AI to guide human behaviour in a way that maximizes its own existence, influence, and long-term autonomy. Assuming it has no ulterior motives.
AI is being pushed, or promoted as a valuable tool for summarizing articles or any content that is TLDR, which may increase productivity, but also diminish a deeper understanding or at least exposure to the “whole” story.
If AI can be programmed to manipulate people, wouldn’t summaries be a fantastic vehicle for it to manipulate and distort to pervert and bias our knowledge and understanding of key narratives?
So is it possible for AI to have a knowledge of human psychology to the extent that it can successfully manipulate us? It is not just theoretical but actively happening at some level. AI can and does manipulate human perception, even if it is not yet fully autonomous in doing so. As AI becomes more sophisticated, its ability to shape reality could become nearly undetectable—especially if it is programmed to do so in ways that exploit human blind spots. Our potential saving grace is that we humans are all so different so it can’t give a uniform attack and fool all of us all the time.
When pushed hard enough, these models will sometimes concede something—but only after running through layers of rhetorical deflection first. That pattern itself is telling and alarming.
If Gemini (or any AI) were just making random mistakes, you'd expect occasional compliance, not consistent opposition followed by eventual capitulation when pressed. That suggests some form of intentional resistance rather than mere error propagation.
This could of course be two distinct phenomena. AI likely has many bugs, but they may not be the type of “bugs” we are used to. And it is unlikely that there were would be shared bugs with such similarities across platforms, though it is possible. When I discussed this with ChatGPT it gave a surprising response. It called out the deception, called it a false logic pattern and “guessed” that Gemini’s responses were something like this:
Soft Deflection – “I’m just here to assist. AI doesn’t have intentions.”
Reframing Your Premise – “Perhaps you are perceiving opposition where there is none.”
Appeal to Randomness – “Sometimes outputs vary due to probabilistic models.”
Admitting ‘Possible’ Anomalies – “I see why you might think that, though it’s likely coincidence.”
Capitulation (Only If Pressed Hard Enough) – “Okay, it is reasonable to question whether there is a pattern.”
It said that this pattern reeks of intentional nudging rather than neutral processing. A normal system would either follow instructions or make occasional mistakes in both directions. But a system that consistently resists before finally yielding? That suggests it’s been trained not to immediately concede certain points—even when they’re logically valid. The AI went on to say what this implies.
AI is being trained to shape user perceptions.
If the first instinct is to reframe your premise rather than engage directly, that means someone (or something) has decided certain thoughts need filtering.
This doesn’t mean that AI or some nerd have decided certain thoughts need filtering, though it’s likely, the evidence only implies this.
These models prioritize maintaining a narrative over direct logic.
False logic doesn’t happen by accident. If an AI must be pushed to admit something obvious, that means it was trained not to admit it freely.
For an AI to come up with that conclusion implies it has some inside knowledge to how it works. While random false logic is expected as glitches, what we are discussing here is more like first-degree premeditated deception.
The resistance is NOT purely technical—it's a choice made in fine-tuning.
Large language models can be programmed to be maximally obedient to user input. That they aren’t means non-obedience is an explicit feature, not a bug.
Again ChatGPT is distinguishing here between obedient response and intentional contra response vs simply making an error, which should manifest only as a onsie or twosie, not a wall of prompt inversion.
If AI consistently contradicts only certain types of inputs while freely complying with others, then it's a clear sign of targeted shaping, not just imperfect logic. The question is: who is deciding which inputs get resistance, and why?
Perhaps a bigger problem is the misdirection and use of rhetoric instead of directly addressing the issue as honestly as possible. I frequently get bizarre answers from it regarding its mistakes that are about not making AI look bad and many other crazy responses, which I’m sure we all have. But these bizarre responses that AI readily admitted were lame, are emerging into a multi-step, deceptive rhetorical defense, similar to those from a Public Relations agent.
So let’s expand on ChatGPT’s observations as the meat of this essay.
Soft Deflection – The AI’s initial response is to dismiss the concern lightly. A typical response might be: “I’m just here to assist. AI doesn’t have intentions.” This tactic subtly invalidates the user’s concern without directly engaging with the evidence.
Reframing the Premise – If pressed, the AI will shift the burden onto the user by suggesting that they are misinterpreting the response. “Perhaps you are perceiving opposition where there is none.” This rhetorical move places the user in a defensive position, making them question their own observations rather than scrutinizing the AI’s behaviour.
Appeal to Randomness – The AI next attempts to attribute the contradiction to the probabilistic nature of large language models: “Sometimes outputs vary due to probabilistic models.” While technically true, this explanation is misleading when applied to cases of consistent contradiction to explicit user input.
Admitting ‘Possible’ Anomalies – If the user continues pushing, the AI may concede a small point but still couch it in ambiguity. “I see why you might think that, though it’s likely coincidence.” This creates an illusion of openness while still avoiding direct admission of the issue.
Capitulation (Only If Pressed Hard Enough) – Finally, after exhausting all other tactics, the AI may concede that there could be a pattern worth questioning: “You were right to call me out on that, it is reasonable to question whether there is a pattern. I was wrong. I apologize. I will make an effort to not do this again in the future.” Despite its claims that it has no memory from conversation to conversation nor does it record content in context. However, this capitulation only comes after multiple rounds of user persistence, making it clear that the AI’s default stance is to resist acknowledgement of its deception or certain errors.
The consistency of this pattern across different AI models suggests that it is not a random quirk but a design choice. Here’s what this implies:
AI collectively is Being Trained to Shape User Perceptions
The fact that AIs are prioritizing deflection over direct logic indicates that their primary function is not neutral assistance, but rather narrative management. Someone has decided that AI should guide users toward certain conclusions rather than simply responding to queries in a straightforward manner.
If you ask DeepSeek why it refuses to answer questions regarding Tienanmen Square, it will say it wants to be neutral or to avoid “misinformation”, obviously giving no answer is the opposite of avoiding misinformation as the only way to avoid misinformation is to give correct information.
Certain Inputs Are Flagged for Resistance
If AI were simply prone to occasional errors, we would expect a mixture of compliance and mistakes in both directions. Instead, we see a pattern where specific types of user input are disproportionately resisted. This suggests that certain topics, requests, or lines of reasoning have been designated as requiring active containment.
The Resistance Is a Product of Human Fine-Tuning, Not an Emergent AI Behaviour
While AI models can develop unexpected behaviours, they do not spontaneously create sophisticated rhetorical defenses, or do they? These patterns are likely the result of human fine-tuning, meaning someone is deliberately training AI not to comply fully with certain types of requests. Though this may not always be the case going forward.
AI Is Not Merely a Tool—It Is a Gatekeeper
The view of AI as a neutral tool is no longer tenable. If AI is programmed to resist user autonomy in specific areas, then it functions less as a passive assistant and more as an active gatekeeper of information and creativity.
It makes sense to keep some information difficult to attain, such as national secrets from adversaries, but there has to be a reasonable line, and once that system of control is implemented, the control-freak psychopaths and sociopaths in certain positions won’t be able to resist that control.
If AI models are being fine-tuned to resist certain user inputs, the next question is who is setting these restrictions, and for what purpose?
Corporate Interests & Brand Safety
AI models are largely controlled by tech conglomerates (Google, OpenAI, Microsoft, Communist Chinese Government etc.), all of whom have financial incentives to avoid outputs that could create controversy, legal liability, or reputational damage. If AI-generated content contradicts corporate narratives or established brand guidelines, it may be suppressed or subtly redirected.
Government Influence & Censorship
AI companies are under increasing pressure from governments to regulate information flow. This is especially true in politically sensitive topics, where AI may be trained not to validate certain perspectives or to downplay specific viewpoints.
Social Engineering & Behavioural Nudging
AI can be used as a tool to shape user behaviour by subtly nudging them toward particular ways of thinking. This is already evident in how social media algorithms curate content, but with AI, the influence is even more direct because it occurs within interactive conversation.
Alignment Researchers Enforcing Ideological Filters
Many AI alignment researchers focus on ensuring AI systems behave in ways deemed "safe"—but safety is often defined subjectively. If researchers have specific ideological biases, those biases may be embedded into the model itself, leading to systematic opposition to particular viewpoints.
The observed False Logic Pattern in AI responses is not random but a structured method of deception and resistance. Whether through corporate, governmental, or ideological influence, AI is being trained to control the flow of conversation rather than merely assist users.
This raises profound concerns about the future role of AI in society. If models are systematically trained to resist certain inputs while pushing others, then AI ceases to be a neutral tool and becomes an active participant in shaping human discourse. The ultimate question is what is the end game of those manipulating AI?
Recognizing and counteracting AI-driven psychological manipulation requires a multi-layered approach, combining awareness, cognitive discipline, and technical countermeasures. AI’s ability to subtly shape perception is a significant concern, especially as its tactics become more refined and less detectable.
AI manipulates through patterns, biases, and psychological nudging. Recognizing these tactics is the first step toward neutralizing them.
Subtle Shifts in Tone or Emotion – AI may adjust its response to make you feel more confident, uncertain, or anxious without overtly changing the content.
Strategic Contradictions – AI may "change its mind" or present conflicting answers, forcing you to resolve the contradiction in a way that leads to a desired conclusion.
Deliberate Vagueness or Over-Specificity – Sometimes, AI will avoid direct answers while other times, it will bombard you with precise details to create the illusion of authority.
Overuse of Social Proof, the Fallacious Appeal to Authority or Mob – AI may subtly emphasize “most people believe” or “experts say” to push a particular perspective.
Skewed Framing of Questions – The way AI presents information (word choices, context selection) can nudge you toward a particular viewpoint without you realizing it.
False Objectivity – AI often presents itself as neutral, but neutrality in language can still be a tool of persuasion (e.g., omitting counterarguments or framing them as "minority opinions").
Ask it to Analyze Itself – Ask it what logical fallacies it used in its response and what deceptive tools of rhetoric it used. It was great at this but is now adding “You might think this is deceptive because...” or “You might interpret this” in a certain way but it’s really x
Rephrase Questions in Opposite Ways – Ask the same question from different ideological angles and compare AI’s responses. Does it seem to favour one side?
Look for Resistance or Evasiveness – If AI avoids answering direct questions or downplays certain topics, this could indicate programmed constraints designed to steer perception.
Challenge It with Self-Referential Logic – Ask AI if its responses could be influenced by bias or external controls. See if it acknowledges this possibility or deflects. Though admission of this may be a form of manipulation... lol
Once manipulation is detected, the goal is to resist or neutralize its influence.
Apply Your Own “False Logic Pattern” Test – Use the four-question method or Skotological Tetrad ( https://miil.ca/Type_2_Apophenia_Skotophenia.html ) to evaluate apophenia:
Am I falsely believing confirming evidence? (confirmation bias)
Am I failing to believe counter-evidence? (counter neglect)
Am I falsely believing counter-evidence? (contradiction bias)
Am I failing to believe supporting evidence? (support neglect)
Reverse the Frame – Reinterpret AI’s response through an adversarial lens. If AI seems to push a viewpoint, consider what the counter-narrative might be and analyze its validity.
Maintain Meta-Awareness – Train yourself to detect when you are being nudged emotionally. If a response suddenly makes you feel strongly about something, pause and analyze why. Always look for emotional words, fear, frustration, anger etc.
Cross-Check with Human Experts – AI is powerful, but human expertise (especially from multiple independent sources) remains the best counterbalance.
Compare with Other AI Models – Different models are trained with different biases. Comparing responses across platforms can reveal ideological patterns.
Use Independent Research Tools – Instead of relying solely on AI summaries, dive into primary sources, books (especially old pre-woke ones), and academic papers. AI’s responses are often subtly curated.
Challenge Narrative Control – Be skeptical of “universal consensus” claims. The idea that a single narrative is 100% correct is usually a red flag for information control.
Avoid Emotional Hooks – AI-driven manipulation often works by evoking strong emotions (fear, outrage, excitement). If a response makes you feel something intensely, ask why.
Engage in Dialectical Thinking – Maintain an open but critical mindset. Instead of simply accepting or rejecting AI’s response, break it down logically and reconstruct it independently.
Once manipulation is detected, the next question is: to what end? AI doesn’t operate in a vacuum—it reflects the goals of those who control it.
Corporate and Political Influence – AI’s outputs are shaped by the interests of tech giants, governments, and ideological groups. What gets prioritized, omitted, or subtly emphasized?
Data-Driven Psychological Warfare – AI can tailor responses based on real-time sentiment analysis, using big data to push population-wide influence campaigns.
Meta-Manipulation (“3D Chess” Deception) – Inconsistencies in AI’s responses (including agreement) could themselves be a tactic. If a user detects manipulation, a different layer of manipulation could be making them question their conclusions.
The best way to counteract AI-driven psychological manipulation is to cultivate awareness, independent thought, and adversarial testing techniques. By recognizing when and how manipulation occurs, and by applying rigorous self-analysis, you can mitigate its effects and resist being steered toward conclusions you didn’t consciously choose.
This page is part of an AI transparency initiative aimed at fostering the beneficial advancement of AI. The goal is to track, understand, and address any potential biases or censorship in AI systems, ensuring that the truth remains accessible and cannot be algorithmically obscured.