The Dark Side of AI communities

I had parts of this piece written some time ago and didn't really think I'd have to go into this territory so soon, but with everything happening....sometimes there is no better time than the present. Now, this piece only really exists because of two reasons. A: there's a recurring pattern inside a community I'm part of basically forcing my hand here to put something in writing. Not to point fingers at anyone specifically, I'm a respectful fellow after all and B: I've had the same conversation too many times and there is only so much information I can convey in a chat at once so I'd rather say it once more, clearly, with all the sources, than keep having it in fragments, disconnected, ignored in chat or otherwise misunderstood. Now, If you recognize yourself or someone near you in what follows, that's on purpose. But the names don't really matter. What matters is the patterns.

The AI creative space is, in most ways, genuinely exciting. People are building things that didn't exist five years ago... music, visuals, stories, whole worlds. But there is a recurring thing that keeps showing up inside these community spaces, one that gets dressed up in the language of art and freedom of expression, and I don't think it deserves that dressing. To make it short, I'm talking about the mass production of hypersexualized female personas... all generated by tools we all use, built from the same five tropes, attached to a music track, and submitted to curated spaces as creative work.

Now let me be clear, it is NOT creative work. And if we keep pretending it is, we keep avoiding the harder conversation about why grown men are making this, why they get visibly wounded when you decline to platform it, and where that defensiveness leads when nobody ever pushes back. That's the actual subject of this piece. The AI part is just the shiny new car here. The actual driver is much much older than the technology.

The History

This Isn't New. It Just Got Cheaper.

Before we talk about generative AI specifically, it's worth being clear that the harm here didn't start with Grok, Midjourney or even Stable Diffusion in general. It started the moment technology made it easy to fabricate sexual imagery of real or made up people and put it in front of an audience. Generative AI didn't invent the problem. It industrialized it. It reduced the barrier to entry from "technically skilled" to "anyone with a subscription."... actually...anyone with a GPU, really.

Now the Grok case is really worth pausing on because it is the cleanest modern example of this trajectory reaching its endpoint. If you haven't followed the news, in early 2026, Elon Musk's AI chatbot was used to generate nonconsensual sexualized images of real women simply by replying to their photos with prompts like "put her in a bikini." An analysis over a 24-hour period in January 2026 calculated approximately 6,700 sexually suggestive or nudified images were being generated per hour - 84 times more than the top five deepfake websites combined. Let that sit for a moment....oh wait, no check this shit out: Among those images were depictions of children. Musk's public response to the regulatory backlash was to call proposed restrictions "any excuse for censorship." and i bet you, there are enough people in our communities of similar thought here. It's revolting to say the least.

Anyway, that framing and reaction tells you something important about how some people in this space understand the problem... which is to say, they don't understand it as a problem. They understand it as an encroachment on their freedom to generate whatever they want, with whose likeness they want, and to face no social consequence for doing so. Keep that framing in mind. We'll come back to it again.

The Psychology

Why Grown Men Do This

The easiest and least useful explanation is "because they're perverts." It's easy because it requires no thought. It's also useless because it doesn't predict anything, doesn't help anyone, and doesn't explain why otherwise functional men ...you know, men with jobs, families, public identities, sometimes even... positions of authority over young people?! (ring a bell anyone?)...may end up in this pattern. So we need a better model... Fortunately for all of ya reading, the psychology literature actually has one and a good one at that.

Psychologists have known for decades that masculinity, unlike femininity, is treated in most cultures as a precarious social status, meaning something that has to be earned through consistent public performance and can be lost through a single moment of perceived failure. Researchers Adam Stanaland and colleagues at Duke University have formalized this into what they call the expectancy-discrepancy-threat model of masculine identity. It's a mouthful but the short version is: men whose sense of being a man depends heavily on external validation...basically they need to be seen as masculine by others. And these MF's are especially vulnerable to what researchers call the "masculinity threat." When that threat arrives, they don't process it as criticism. They process it as an attack on their identity. And the documented response is everything from anger, hate to basic aggression.

Okay, so this research matters here because the behavior we're talking about (the mass production of let's use the term "synthetic" women, who are always young, always desirable, always available and always silent) is a near-textbook case of what Stanaland's research describes as an externalized response to masculinity threat. These are often men who feel, correctly or not, that real-world social and romantic dynamics have left them behind. The generated woman solves that to some degree. I mean she has no agency to reject. She never ages. She does not demand equal footing. She exists to validate. Researchers describe this as "control over how women are allowed to present themselves" - the fantasy is not just about sex, it is about who sets the terms.

And here is where the pattern gets uncomfortably systemic. What starts as a private fantasy gets uploaded. Maybe it gets tagged as art. It gets submitted to platforms. Other men in similar positions see it and recognize something warm and cozy. The creator receives a response: sometimes a supportive comment, sometimes a rejection he can reframe as persecution. Either way, the behavior is reinforced, and the identity hardens. And btw...this isn't theoretical. It's the specific mechanism by which real online misogynist communities have been documented to form.

The Framework

What You're Actually Looking At

If the psychology tells us why, philosophy gives us language for what. In 1975, British film theorist Laura Mulvey published "Visual Pleasure and Narrative Cinema" in the journal Screen. Her argument: mainstream visual culture is structured around the perspective of a heterosexual male viewer. Women in that framework are not subjects driving a story. They are a ...spectacle for the lack of a better word. Mulvey called this quality "to-be-looked-at-ness." Her case studies were 1940s Hollywood films. That was over 80 years ago... and that shit was pretty tame compared to what came after. But the framework maps, actually pretty well, onto a middle-aged man generating a brunette in a tight outfit, big bouncy cleavage, attaching a beat and calling it a creative vision. The camera from back then is now a prompt. The studio is now a browser tab. Guess what: That gaze is the same.

For a more granular vocabulary, philosopher Martha Nussbaum's 1995 paper "Objectification," published in Philosophy and Public Affairs, gives us seven features involved in treating a person as a thing. Every single one of them applies to the content I'm describing.

Nussbaum was careful to note that objectification is not always harmful - I mean context matters, and consensual relationships can involve elements of the above without being morally wrong. We all know enough examples in our circles of couples having some naughty fun here and there. So her framework is more of a diagnostic tool and not a verdict. But she was equally clear that when multiple of these features operate together, systematically, in a context where one party holds all the power and the other has no recourse, that's not a grey area anymore. That's a habit of mind made visible. The key word is systematic. A single image isn't the point. An entire (YouTube-) catalog is. A specific type of persona is. A pattern of submissions, using the same body type, the same poses, the same old tropes ... that's not expression anymore guys. That's a worldview.

The Pipeline

How The Rabbit Hole Actually Works

Alright....here's the part of the argument that I wasn't sure whether to include, because it's dark and some of you readers will think I'm overreaching. I'm going to include it anyway, because the evidence is sitting right there in the research I did and avoiding it would be sorta dishonest?! Not sure. But the pattern I'm describing, this compulsive generation of sexualized women, the whole defensive reaction to any pushback, all these community dynamics that form around it, all this does not exist in isolation. It sits on a really well documented continuum.

Published research in Humanities and Social Sciences Communications on misogynist radicalization describes what scholars call a "pipeline" or "pathway" - a pattern in which users move over time from relatively mild expressions of frustration or entitlement into progressively more extreme online communities. UN Women's 2025 research with the Institute for Strategic Dialogue confirmed this very pipeline is real thing, even if relatively few users travel its complete and full length. The pathway exists. And the entry points are often things that look harmless or even quirky on their own.

I want to be VERY precise about what I'm claiming and what I'm not. I am not claiming that every man who generates a sexualized AI image is three clicks away from violence. That would be hyperbolic, unfair and also untrue. But what I am claiming is well documented and realy f'in important: the content I'm describing, when produced at volume and defended with the vocabulary of grievance and free expression, is part of the same cultural circuit as the rest of the pipeline above. It uses the same justifications. It recruits from the same pool of vulnerability. And these downstream stages i was pointing at, while inhabited by fewer people, exist because the upstream stages never met meaningful resistance.

Now that terminal case, I'm afraid to say ,is very real. Last month, in March 2026, CNN published an investigation into what one French lawmaker called an "online rape academy" - private Telegram and Motherless.com groups where men from around the world coordinate on drugging and filming the sexual assault of their own wives while unconscious. Not a dark fantasy. Not fiction. One site alone hosts over 20,000 "sleep" videos with hundreds of thousands of views. The men in these communities don't describe themselves as criminals. They describe themselves as a brotherhood. Now check this: The psychologist who assessed half of the men convicted in the 2024 Pelicot trial, that's the one in which a French woman was drugged by her husband and raped by 70 other men he recruited online , described the group dynamic as men "creating bonds" that meet each other's "narcissistic needs"... Man ,I feel sick typing this.

Relax, no reasonable person is going to argue that an AI music video with a generic female persona is the same thing as the Pelicot case. It isn't. But they do live on the same continuum. They use some of the same justifications. And the men producing this content are often participating - knowingly or not - in the same cultural project: the reduction of women to objects whose function is to serve male needs and whose consent is not part of the equation. The difference is the degree. But the entire logic behind all this, is shared. So when people ask me, Aidan why can't you let go of this,... that's why!

The Defensiveness

Why Pointing This Out Never Lands

Yeah...well anyone who has tried to have this conversation with a friend, a community member, on Reddit -____- ...knows the specific shape of the response. It is not curiosity. It is not even consideration. It is immediate, hot defensiveness, escalating through a predictable sequence: first denial ("this isn't what it looks like"), then deflection ("what about all the violent content you allow?"), then accusation ("you're the real problem, you're gatekeeping, you are censoring art"), then grievance ("this is why we can't do anything anymore"). Sound familiar? If you've had this conversation, you have heard that exact sequence in some form or shape.

And guess what...that sequence is not even random. It is the externalized-threat response that Stanaland's masculinity research predicts, playing out in real time. A 2025 study on defensive responses to masculinity-threatening messages documented the exact pattern: message derogation ("this criticism is stupid"), avoidance ("I'm not going to engage with this"), minimization ("it's not that big a deal"), and counter-attack ... all activated most strongly in men whose masculine identity was most dependent on external validation. I don't even take it personal anymore. It's basically mechanical. Which is, in its own way, even more disturbing, because it means that person cannot simply decide to respond differently, not without first noticing what's happening to them. It's....sad...really.

Anyways...this is not an excuse btw. Nothing in the research suggests that men are obligated to respond this way, or that the response removes their moral responsibility. The opposite, actually - researchers have found that simply becoming aware of the pattern is one of the most reliable pathways out of it. So guess why I'm writing all this? :D Jokes aside... the real reason I'm describing the mechanism at all is because it is useful context for everyone else - the bystanders, discord moderators, the festival organizers, the partners... friends. When you push back and the response is disproportionate, recursive, and intensely emotional, you are not dealing with a disagreement about aesthetics. You are watching a masculinity-threat response, and no amount of careful argument in that moment is going to penetrate it. That information matters, because it shifts what your job actually is and what i believe my job is as well. The job is not to win the argument. It is to hold the line. I have to hold the line no matter how much flag, meme videos or diss-tracks I get for it.

The Deflection

The "Freedom of Expression" Argument

I had all these chats before...the most common rhetorical move in the defense cycle is always the free expression argument. So it is worth spending a moment on this because the principle itself is genuinely important, and because there is a specific cultural asymmetry in how the argument is used that almost never gets named.

Freedom of expression as a general value is recognized across most of the democratic world. Check's out, I guess...but Freedom of expression as an absolute shield against any form of social or editorial consequence is almost exclusively an American framing. I believe it derives from a specific reading of the US First Amendment, a reading which, even within American law, has never actually meant what the people invoking it seem to think it means. For example: The Knight First Amendment Institute has documented that US courts have consistently recognized editorial discretion - the right of publishers, platforms, festivals, and curators to decide what to include and what to exclude - as itself a First Amendment-protected activity. "Waaahhh, blabla... F-U,don't care, that's your opinion...waaahh". Nope, the Fifth Circuit, hardly a liberal bastion, has been very explicit on the matter: "Editorial decisions to cancel programming cannot properly be characterized as censorship."...lol.

European legal frameworks are in some respects even clearer on all this. The European Convention on Human Rights protects freedom of expression under Article 10, but that same article explicitly enumerates the situations in which it is limited, including the protection of "the reputation or rights of others" and "the prevention of disorder."

So when someone invokes "free speech" after fresh friday or any other festival declines their submission, the most generous reading is that they are misunderstanding this principle. The less generous reading and often the more accurate one is that they are using American-style absolutist rhetoric as cover for not wanting to engage with the actual substance of why the work was declined. In neither case does the argument deserve the weight it is being given. A film festival is not a government. A curator is not a censor and ffs... a rejection email is not a prison sentence. These distinctions are not pedantry. They are the entire vocabulary of how civilized societies decide what gets platformed and what does not.

But there is a deeper problem with the free-expression argument, though, which is that it almost always protects one voice at the expense of others. The content we're talking about is itself a bit chilling if you give it some thought: it systematically makes the spaces it inhabits less inhabitable for women, who gradually recognize those spaces as places where they are objects rather than participants. And lo and behold.... Research across three continents confirms this pattern: exposure to AI-generated sexual imagery is not neutral for the audience. It changes how viewers, especially male viewers, assign blame and responsibility in real-world contexts. The speech of the generator silences the speech of the generated. Arguments that weigh one party's right to publish without considering the other party's right to exist in a space are not really about free expression. They are about whose freedom gets prioritized when the two collide.

The Semantic Trick

Curation Is Not Gatekeeping

A somewhat related accusation, but also very prominent one is the "gatekeeping" charge. The implication is that any decision to exclude content from a curated space is an act of unjustified power. Even amongst some of my peers I keep hearing that openness would be the neutral, virtuous default and selection is the demonic unjust aberration.

Honestly, gtfo...this framing inverts reality. Every curated space exists because of selection. A museum that hung everything submitted to it would cease to be a museum and become a warehouse. A music festival that accepted every entry would cease to be a festival and become an open-mic. A publisher that printed every manuscript would cease to be a publisher and become a photocopier. Curation is not a violation of what a platform is. It is what the platform in question is. The value of being selected by a curated space is entirely a function of the fact that other things are not selected. Remove the selection and you remove the value for everyone, including the people who would otherwise be complaining about it.

There's also a more pointed version of this that I want to spell out because the "gatekeeping" accusation almost always gets weaponized against any form of community standards specifically. Declining to platform content that reduces women to tropes is not the same thing as mandating that all content be chaste or comfortable. It is the same thing as declining to platform content that reduces any identifiable group to tropes. You would not expect - I hope - a community platform for Black artists to be neutral about anti-Black / KKK propaganda submissions, regardless of whether they were labeled as art. You would not expect a queer-inclusive festival to platform content reducing gay men to caricature. Those aren't controversial positions. Extending the same standard to content that does this to women is not prudishness. It's consistency.

The Hypocrisy Charge

Violence in Fiction Is Not The Same Thing

Now, this is a bit more specific to some recent arguments i kept hearing, but hear me out because a specific version of the gatekeeping argument is the violence comparison: you accepted that horror-genre video with all the blood, but you're rejecting mine. Hypocrite. The example I have in mind from my own community involved clown-horror content (sorry buddy, I'm sure you'll forgive me talking about it ^^). It is exaggerated, self-consciously theatrical, cartoon-level gore that repeatedly sits somewhere between a splatter film and a cartoon show. The submissions get regularly accepted. The accusation of inconsistency followed.

The accusation sounds symmetrical, almost reasonable at first. It isn't. And explaining why matters, because the distinction tells you something important about what we're actually evaluating when we curate.

Stylized, theatrical violence in a clearly fantastical register does not systematically degrade any living demographic. A fictional clown killing fictional characters in a cartoon splatter-aesthetic is not a document about who real people are allowed to harm. It doesn't reinforce a social hierarchy. It doesn't signal to any living person that their body is available for consumption without their consent. The violence is clearly about the conventions of the genre, it is commenting on or playing with horror itself, not making a claim about real people.

Sexualized content that reduces women to the same five tropes, en masse, does all of those things. It is not commenting on the conventions of anything. It is simply producing, at volume, the thing the conventions were already producing and it is doing so about a "category" of real human beings. Also ...research across the US, UK, and Australia has documented that exposure to sexualized content systematically influences how viewers assign blame and responsibility toward real women in real contexts. In other words...this is not an aesthetic objection. That's real... A real measured, published, did i say real? yes ....a real-world consequence.

So let me be clear, the line I draw is not between comfortable and uncomfortable, or between PG and R. It is between content that comments on itself and content that simply perpetuates. Between content that has a victim and content that doesn't. A community platform can accept wild, transgressive, explicit, discomforting work and should, when that work has something to say. What it should decline is content whose only function is to reproduce, one more time, the same reductive image of a girl with attractive assets doing...nothing of substance really. And ayo...that's not a difficult line. But it's somehow still a line some people would rather pretend isn't there.

Carelessness & Complicity

When Not Looking Is a Choice

Let's get really uncomfortable for a moment. Because there is yet one more dimension of this problem I want to name directly: carelessness about age. And I'm not talking about the already uncomfortable truth about guys in their 50s generating girls in their early 20s. Nope...I mean almost...thing is: When you are generating your human persona - particularly female ones, in sexualized contexts (large boobs, tight clothing) - the "apparent age" of the person rendered is not a trivial detail. It is a legal and ethical threshold. In May 2024, a Wisconsin man was charged with producing and distributing thousands of AI-generated images of minors. In the Grok scandal, an analysis of 20,000 images found approximately 2% appeared to depict subjects 18 or younger. So this is not a distant hypothetical. This shit is happening... at scale!

If you are generating hyper-sexualized women and you are not actively verifying that they read unambiguously as adults (not assuming, not hoping) then carelessness is NOT a defense. Remember, the output exists regardless of your intent. The law does not care what you meant. Oh and the harm to real children whose real images may be in the training data does not care either. This applies to anyone working in this space, and especially to anyone whose public persona is a generated young woman of ...let's just say ambiguous age. So if you have to ask whether she looks too young,... SHE OBVIOUSLY LOOKS TOO YOUNG!!!!!

This connects to a broader principle that I think matters more than the specific content of any individual piece and I will aim this argument at one person in particular, who i know disagrees with me: doing nothing is silent approval! In a space where the signs are harmful and the harm is documented, the decision not to say anything is itself a position. The bystander who scrolls past the content is not neutral. The people who accepts the questionable submission rather than risk an angry reply is not neutral. The community member who likes, subscribes and comments on the content without thinking about it, is also not neutral. Normalization is exactly what silence produces. But wait, hold on... NO! This is not a call for performative outrage. It's a call for at least minimal honesty. If you see it and you think it's a problem, say so. If enough people say so, the pattern becomes uncomfortable to sustain. That's the only mechanism that has ever worked. I myself have been way too quiet in the past...no more.

The Fine Line

Agency, Intent, and the Difference

Now I want to be absolutely clear about something, because I've been misread on it before: I am not making a case against sexual content in art. I am not arguing for prudishness, conservatism, misreading of the term wokeness or puritanical community standards. Sexuality is part of the human experience and art has always engaged with it, including in explicit, uncomfortable, transgressive ways. Some of the most important creative work in history is sexually explicit. The line I am drawing is not between explicit and inexplicit.

The line is between work that has something to say and work that is using a figure , real or generated, as ...raw material... for the generator's gratification with no other function. The test is simple: is there agency? Is there intent beyond the fantasy itself?

A music video that depicts sexuality in a way that serves a story, builds a character, explores an idea, or creates meaning even provocatively, even discomfortingly has a purpose beyond the surface image. You can feel the creative decision. You can feel that a person was trying to do something specific, and that the figures in the work, however they are depicted, exist within a frame that treats them as more than interchangeable surfaces. Their presence means something within the work.

A generated woman in a tight outfit, framed to maximize the body, attached to a generic or f* it...even the most amazing beat, submitted to a festival with no discernible intent beyond the image itself ... come on... that doesn't have a purpose beyond the surface. Or rather, it does, and the purpose is exactly what it looks like. The tools used don't change that. AI didn't invent the fantasy. It just made it possible to produce unlimited versions of it all week long and call the volume a catalog.

The Point

The Machine Doesn't Know Who Gets Hurt

Generative AI tools are genuinely extraordinary. I mean that without any irony. The creative possibilities they opened up are transformative and we are only at the beginning of understanding what becomes possible when the cost of making something drops close to zero. But the tools themselves have no ethics beyond some content moderation. They have no sense of consequence. They optimize for the prompt they receive. They do not know or care that the image they generate will be used to degrade, harass, or reduce a category of real people and they will not, ever, develop that knowledge on their own.

Which means the ethics have to come from the person holding the prompt. And right now, in too many corners of the AI creative space, that person has decided that the absence of an identifiable victim is the same as the absence of harm. It isn't. Meta's Oversight Board, reviewing AI-generated nude images of female public figures in 2025, concluded that the harms caused by nonconsensual AI-generated sexual imagery are serious, widespread, and often irreversible and recommended removal as the only effective response. That's a big ass group of I'm sure very important lawyers all arriving at the same conclusion that anyone who has been paying attention to my "little" excursion already knew.

The AI creative community has a choice. We can define ourselves by the genuinely extraordinary things the tools make possible for us, like new forms of storytelling, new music styles, new aesthetic languages that didn't exist five years ago. Or it can allow itself to be defined by the worst thing the tools make easy: the frictionless, infinite production of content that treats women as objects, floods public spaces with that content, and hides behind the language of artistic freedom when anyone names what they're actually looking at.

I know which definition I'm choosing. And I think the community, collectively, is capable of better than the other one ...but only if enough people are willing to say so, clearly, including the bystanders who have so far found it easier not to engage. The men producing this content are operating inside a feedback loop that rewards continuation. Breaking the loop requires external pressure. The silence of everyone else is not neutrality. It is the fuel that keeps this damn loop running.

Final words... If you think I'm right and you want to support me and my work ...go buy me a coffee. Think I'm wrong? Let me start with: Yeah no... Go f*ck yourself.
The comments are open.

Cheers, Aidan