The Algorithm Will See You Now: AI and the Future of Adult Entertainment
Everyone wants to talk about AI replacing sex workers. The discourse is predictable: Will AI make human performers obsolete? What happens when you can generate your ideal woman with a text prompt? These aren’t shallow questions as much as incomplete ones, asked by people who fundamentally misunderstand what this industry has always been selling.
What I find interesting about the AI conversation in adult entertainment is that it assumes the product is visual and auditory stimulation combined with a basic pandering to fetishes or conversational interaction. And that’s definitely part of it. But if that were the whole story, the internet’s (for all intents and purposes) infinite supply of free porn, including animated and 3D content that’s already divorced from human limitations, would have made paid adult content obsolete years ago.
Let me walk you through why the future of this industry is messier and more interesting than “hot bot replaces hot woman,” and what happens when technology finally forces a distinction between content and connection.
Where AI Looks Like Liberation & Efficiency
At face value, for current creators, there are genuinely empowering use cases for AI in adult content creation, and pretending otherwise is intellectually dishonest. But I say *current* creators for a specific & important reason; one I’ll circle back to later.
Consider the woman who’s “aging out” (I don’t believe in this term but many consumers do), pregnant, or injured. AI offers a way to capture yourself at your peak and monetize that image indefinitely, leaving you to be at the park with your kids, be recovering from a surgery, or simply on vacation, makeup-free, and still earning. You can avoid the physical toll & risk of cosmetic procedures, the expense of elaborate costumes and sets and the cleanup after a messy shoot involving oil, all whilst being & having whatever your fans want in any given moment, be that 32K breasts, a waist to hip ratio that would require a medical tourism visit to Brazil to achieve, or a new latex catsuit & access to a swimming pool.
Want to do boy-girl content without the infection risk? AI can generate it. Want to be a redhead for a client who loves everything about you except your blonde hair? Quick fix. Guy in your DMs wants a video of your red toenails but yours are currently white? Not a problem. You can scale your wardrobe, your locations and your appearance infinitely without the linear constraints of having an actual body.
There’s an alluring potential here for autonomy, efficiency and power.
But there are three issues I see with this usage:
i) This only works well for creators who’ve already established an identity in this industry. AI can replicate your image, but it can’t create your brand from scratch. The woman who benefits most from AI-generated content is the one who already has an audience that knows her, trusts her, and wants her specifically.
A new creator trying to enter the market with purely AI-generated content is competing with every other AI generator on the planet, most of whom are men. There’s no scarcity, no story, and no reason to care about her. These are just pixels that came from anywhere, that any narrative can be attached to (or indeed, switched up) in an instant. And whilst that will work for some consumers, it will fall flat with many more, in my opinion.
ii) We have no clue what setting expectations for sexuality unrealistically high might do to genuine human connection. Actually - I do have some idea, based on the feedback from some fans, and it’s not pretty. These men claim that they have “PIED” from excessive consumption of unrealistic explicit material1. They claim they cannot become aroused for the nuance that underpins real interactions, only oversaturated Picture Music Video style stimulation pumped at rapid speed. I also have witnessed first-hand men being shocked that most women have cellulite, toe hair, and that “it” is not usually pink, leaving women to try lotions & body brushes, toe waxes and vulva bleaches in an attempt to “keep up”.
What happens when we skew expectations to the truly impossible to achieve? Do men become entirely unaroused by the physical female form? Or does it become even more sacred and valuable? This unlocks a broader philosophical question: Should we cater to every sexual desire? Or is there value, for humanity’s sake, in keeping some things unachievable and having a little resistance baked into fantasy?
iii) For many adult entertainment consumers and even some fetishists, the fantasy isn’t rooted in the purely physical, but in what it represents.
Take bimbofication for example. The kink isn’t just about big lips and bigger breasts—it’s about extremity. The transgressive thrill of subverting classical feminine ideals (intelligence, demureness) with an irreversible commitment to the aesthetic. It’s the “no going back” aspect of actual surgical modification, the rebellious celebration of supposed ditzy-ness as choice rather than accident. AI can generate the visual elements of this, but it’s a hollow representation - the physical features without the sacrifice, commitment, and the story of transformation. I’m not convinced that lands the same way.
Similarly, those into Muscle Worship typically love the look of a muscular woman, sure, but for many it’s just as much about the admiration of (again) the sacrifice, the hard work, and the rarity it took to get to that place. Men into this kink often love imagining a real woman out there, physically strong enough to take him down, showing off strength she took years to build. They want to know how much more you can lift than them, how big exactly your biceps are. AI cannot feasibly sate these questions with answers.
Even traditional beauty such as white teeth, glowing skin, and glossy hair, is appealing partly because it’s hard to obtain. It’s rare, it’s special and men don’t see it every day in real life. That scarcity creates value. I don’t think AI solves for this preference for the genuinely rare and difficult to achieve when you know it’s not real. (What happens when you can’t tell what’s real and what’s not will have to be a future conversation).
The one exception might be outfit fetishes. Latex, nylons, satin, high heels, etc. often root their appeal in symbolic representation rather than human achievement. Heels clacking down a hallway signal dominance. Nylons might evoke memories of an older woman from someone’s formative years. These fantasies are more about thematic association than individual accomplishment, and novelty in this realm often matters more than authenticity, in my experience.
For pure outfit fetishes, AI probably works fine. It can generate endless variations of the aesthetic without the representation problem that plagues body-based kinks.
But that’s a pretty narrow slice of the market, and not one that would be mutually exclusive for current creators to capitalize on with AI.
So yes, AI offers efficiency and apparent autonomy. But efficiency for whom? And what happens when everyone has access to the same tools?
Where AI Fails: The Limits of Algorithmic Desire
The Authenticity Paradox
A question I keep returning to is: Does humiliation work if it comes from an algorithm?
In my niche—online domination—the entire psychological fantasy structure depends on the sub believing that I, a human being with judgment and contempt and genuine opinions, find them pathetic. The degradation has weight because it comes from a person who has chosen to (or simply cannot help but) feel that way about them.
But AI doesn’t feel anything. It doesn’t judge you. It doesn’t find you pathetic or impressive or boring. It generates text strings based on pattern matching and probability distributions.
Can an AI chatbot roleplay humiliation? Absolutely. Current models can already produce technically competent degradation scripts (kind of; they are very bland and obviously robotic, in my opinion). But does it land the same way when you know—and you do know, even if you’re pretending otherwise—that there’s no actual disgust or mockery or rejection behind it? That it’s just math?
I suspect not. Or at least, not for the clients who were paying for the real thing. Maybe there’s an audience that’s fine with algorithmic degradation. But that’s a different market, serving a different need: fantasy-as-entertainment rather than fantasy-as-encounter.
The Calibration Problem
Even if we solve for authenticity, perhaps through increasingly sophisticated simulations of emotion, or maybe through users who simply don’t care, there’s still a calibration issue.
AI models are trained to be helpful, almost sickeningly placating. They’re designed to please users, to avoid causing harm, and to stay within safety boundaries. This creates a fundamental tension in a context where the entire point is to push boundaries in ways that feel genuinely risky and exciting. It also creates some humorous possibilities in the context of consensual humiliation: “I’m a worthless loser who should never date again!”, “Aww, Steve, it sounds like you’re feeling pretty low. Here are some numbers to local helplines…”. Good luck getting off to that. And don’t even get me started on CBT. The only way you’re getting a task in that fetish from current LLMs is to ask it how NOT to hurt your balls and do the opposite.
Yes, there are “uncensored” models and jailbreaks specifically for sexual content. But they still lack the human capacity to read between the lines. An example might be distinguishing between actual consent and compliance born from not wanting to disappoint. Or hearing a fan reveal that he’s a virgin and that we can explore that kink if I want, but seeing the deep-seated shame in his eyes as he speaks this truth, which tells me that’s absolutely not a good idea for whimsical fetish play.
This isn’t a bug AI can (or will) fix with better training data. It’s a feature of human interaction to have the ability to pick up on subtle cues, to feel into the emotional truth beneath the words, and to make judgment calls that balance desire against wellbeing.
Can AI learn this eventually? Maybe. But we’re not there yet, and the learning curve probably involves ethical nightmares that no LLM realistically wants to touch if it doesn’t have to.
The Technical Limitations
And then there are the hilariously mundane ways AI just... doesn’t work yet.
As things stand, from personal experience, it can’t even generate basic sexy tasks for fans without extensive hand-holding that made AI-assistance prohibitive, and I’m reluctant to believe this was just user error; I know how to prompt. I’ve been experimenting with this for the better part of a year, and even models supposedly designed for adult purposes failed laughably, providing instructions that either weren’t instructions at all (e.g., “Sink into a submissive space. Breathe in, breathe out. You’re Mine” - as the entire “unique submissive task lasting 10 minutes”) or made no sense. The latter was the most common error, even with clarification of prompting and calling out. We’re talking contradictory instructions. Games which ask you to roll dice and then provide rules based on a coin you were never told to flip. Tasks with no ending, no escalation, or nonsense tedium. No submissive man wants to have to teach his domme how to dominate, trust me. And no one seeking JOI wants to have to write the instructions themselves.
And visually; the physics of bodies in motion are wrong in ways that trigger uncanny valley responses, and the content starts to have a sameness to it—you can spot AI-generated porn pretty quickly once you know what to look for.
These will almost certainly get solved. But by the time they do, we’ll have new problems as mentioned below: deepfakes of real people used without consent, and synthetic content that looks underage. An arms race between detection and generation that nobody wins.
The technology will definitely get better at producing content. It’s not yet clear that it will get better at producing connection.
The Split Industry
My hypothesis is that AI will cleave this industry in two.
Category One: The Objectified Beauty Market
This is the segment AI can serve, and serve well - for a certain audience. If your value proposition is primarily visual novelty - new outfits, new scenarios, new aesthetic combinations; you’re competing in a market that AI is about to flood.
The economics here get ugly fast. When the barrier to entry drops to zero (anyone with image generation software becomes a “creator”), supply explodes whilst demand remains relatively fixed. Prices will collapse, and the market will get saturated with content that’s visually flawless (in time; we still gotta fix hands & tattoos), endlessly customizable, and costs nearly nothing to produce beyond the initial setup.
This isn’t even theoretical at this point. We’re watching it happen in real-time with tools like jailbroken Pika & Runway Gen-3, where men generate hyper-specific content tailored to niche fetishes, all without involving actual human women. Because the human element doesn’t really matter for this segment of the market. In fact, the most popular OF models often succeed with incredibly banal engagement - “say hi or blocked,” “me or a PS5?”, “taking boyfriend applications”, followed up with male chatters from all over the world upselling fools in the DMs. This is formulaic, low-effort work that could easily be replicated by an AI persona. If that’s the level of interaction customers are satisfied with, then an AI-generated egirl with a fabricated backstory (a shy virgin who loves astrology & kawaii culture, whatever) will serve that audience just fine because they’re looking for the aesthetic of accessibility. And AI can deliver that.
This results in a race to the bottom where human creators can either compete on price for this type of mass content & access (unsustainable) or get pushed into producing increasingly extreme content to differentiate themselves (unsafe).
Who wins in this category? Probably no one, at least not long-term. It becomes a commodified wasteland where the only winners are the platform owners taking transaction fees on AI-generated content. Here I think we’ll see subscription based models for users to customize physicality, scenes, outfits, length of videos or games, etc. Rudimentary, non-AI versions of these types of gamified content already exist (FapRoulette, EdgeMePlease, etc.), and AI will only make visual and auditory elements more variable.
And let’s be clear: as a model, you’re not going to be competing with your peers here. The same technology that lets YOU as a model “take back your power” also lets any man with $100 and a GPU generate content that competes with yours. It lets him flood the market with “gooner” AI content—infinite variations of increasingly extreme imagery designed to trigger compulsive consumption—one-sided fantasy without the complication of another person’s presence.
Which brings me to what actually survives:
Category Two: The Human Connection Market
The men drawn to purely AI-generated content were never looking for exchange or collaboration in any real sense. This becomes more a gray area with AI girlfriends, which I’ll touch on later. But when it comes to content consumption alone, they have wanted unilateral consumption without boundaries, without someone else’s humanity complicating the fantasy. That’s a real need for fetishists looking for fetish dispensers, and AI serves it. Ironically, it’ll be a circle-jerk of content for men created by men, but when you’re just looking for a box to be checked, I suppose that doesn’t matter.
But box-checking and fetish-dispensing is not the whole story of what this industry has been primarily selling. It’s the adult *entertainment* industry, not the adult *content* industry. And as humans (social creatures), we’re not built to only entertain ourselves. You could generate AI-porn yourself, but do you want to? You could cook for yourself but you still enjoy going out to a restaurant to eat. You can make yourself laugh but it’s more fun to watch stand-up comedy. Some things work better when there’s an “other” involved; when there’s genuine exchange rather than just consumption.
So yeah; AI can generate a hot video of a woman in latex. But it cannot generate twenty years of conscious experience reading power dynamics. It can produce a script for verbal humiliation based on 20 years of data in this realm. But it cannot calibrate in real-time when someone’s fantasy is bumping up against their actual psychological limits. It can create the aesthetic and thematic representation of dominance. But it cannot hold the weight of someone’s secrets, their shame, their desperate need to be seen and degraded and cared for simultaneously.
In my opinion, the human women who survive and thrive going forwards will be the ones who were never selling bodies to begin with. It will be those who were selling understanding, and most importantly; CONTEXT. The alchemy that happens when someone feels truly known in their desire, when fantasy stops being a product you consume and becomes a space you inhabit with another human being.
This is the market AI struggles with. Not because the technology can’t improve (it will), but because the value isn’t in the output—it’s in the exchange itself.
Economics kind of proves my point here: if what clients wanted was just visual stimulation, piracy would have destroyed this industry decades ago. There’s infinite free content online. Always has been. And yet here we are, with parasocial relationships intensifying, with men paying thousands for custom content, for the girlfriend experience, and for domination that feels personal and dangerous and real. The pixels are the medium for myriad reasons (anonymity, distance, accessibility), but they are not the prize. These men have been paying for *presence*, and there’s a part of them that wants, or perhaps even *needs* to be perceived in the vulnerability and intimacy of that context.
The men who don’t want or need this, and who perhaps even enjoy the exploitative nature of some forms of sex work…well; they can use AI going forward, maybe even for finding a girlfriend…
The AI Girlfriend Question
But what about AI companions? The chatbots designed specifically for emotional and romantic interaction?
This is where things get murkier. In a recent interview, Sam Altman observed that 'there are more people that want [a deep connection with an AI] at the current level of model capability than I thought' (as cited in Kantrowitz, 2025)2, and the market is responding. Replika has millions of users. Character.AI is popular. There’s clearly demand for parasocial relationships with AI that goes beyond pure content consumption.
These tools serve a different function than traditional adult entertainment. They’re filling a void for people who want consistent emotional availability without the messiness of actual human needs and boundaries. Research on human-AI relationships suggests these interactions can provide genuine emotional support for people struggling with loneliness or social anxiety3. They’re therapy-but-make-it-sexy for people who can’t or won’t navigate real relationships.
But is this what adult content consumers are seeking? I argue no, and where they are, I personally always shut it down, and encourage other creators to do the same. This boundary-less convenience is impossible for a human to replicate, and rightfully so. Co-dependence in this sense should not be encouraged in my opinion, and evolutionarily, it’s not even how men - hunters, read: chasers - are naturally wired. My clients are turned OFF when I am chronically online. I have been using the phrase “make yourself available but never TOO available” in my mentoring of peers for decades. What’s the thrill in engaging when there is no scarcity (e.g., value) to the proposition? When the fear of missing out doesn’t exist? How hot am I to you (especially as a dominatrix) if you know I’ll always message you back within seconds with something I think you want to hear? Not hot at all; not energetically, and not mathematically - with variable ratio training being the most addictive behavioral psychology conditioning model.
My clients aren’t buying convenience anywhere as much as the layperson might believe. They’re actually buying a thrill, dotted with a little risk. The thrill of the possibility of genuine connection, the rush from the realness of another person’s disgust or strategic indifference, and the interaction with another person with agency and boundaries, who might tell them no.
You cannot get that from a chatbot, no matter how sophisticated the model. And this creates an interesting economic question: if AI floods the market with cheap, accessible fantasy-content and companionship, does it devalue human performers? Or does it create a premium tier for people who specifically want the human element?
I suspect the latter. We’ve seen this pattern before with automation in other industries. Mass production doesn’t eliminate craft goods—it creates two markets with different value propositions and price points.
AI won’t destroy sex work. It will split it. The creators who’ve been doing this work with intentionality, who understand they’re selling connection rather than content, will be fine. Better than fine, probably, once the bottom tier gets automated away and what’s left is work requiring actual humans.
Who Decides? Regulation and Power
So who gets to decide what AI can and cannot do in this space anyway? Right now, the answer is “whoever owns the platform,” which is a polite way of saying “tech companies with a financial interest in maximizing engagement while covering their asses legally.”
The result is a complete mess:
OpenAI won’t let you use ChatGPT for sexual content, even consensual adult scenarios.
Character.AI implemented strict filters after public pressure about underage-presenting characters.
Uncensored models are flourishing in gray markets with zero oversight.
Grok lets users generate deepfakes of real people, then claims platform immunity when called out.
Clip sites claim that AI models can only exist as representations of the real human form managing the account, then don’t enforce that.
There’s no coherent framework governing any of this. Just a patchwork of corporate policies, reactive moderation when something blows up publicly, and jurisdictional confusion about what laws even apply.
And of course the people most affected by these decisions? Sex workers whose images are being used without consent, whose labor is being devalued by algorithmic competition? We have almost no voice in how these policies get written.
This isn’t new. Sex workers have always been shut out of conversations about our own work conditions. But AI raises the stakes because now we’re fighting for image rights and the basic ability to consent to how our faces and bodies appear in content we didn’t create.
Though maybe that’s changing. This is no longer just a sex worker problem. Taylor Swift has had deepfake porn made of her. Grok’s “undress” feature works on literally anyone’s photo. When the issue affects celebrities and regular civilians instead of just a marginalized population, maybe people will actually care enough to do something about it.
The deepfake issue deserves its own essay, but here’s the reality: if someone generates AI porn using your image, there’s currently almost no legal recourse. Some jurisdictions are starting to address this, but enforcement is nearly impossible when platforms hide behind “we can’t police user-generated content at scale.”
Then there’s the age-appearance problem. If AI generates images that look underage but aren’t of any real person, is that legal? Harmful? The legal frameworks aren’t settled, and different countries will answer very differently.
What I know is this: the people making decisions about how AI gets deployed in adult entertainment don’t do this work and don’t understand the power dynamics at play. And in many cases, they actively benefit from keeping sex workers marginalized and without a platform to push back.
If we actually want regulation that protects workers while still allowing for technological innovation, sex workers need to be at the table. Not as tokens or representatives of some monolithic “sex worker perspective,” but as the actual experts on what consent looks like in practice, what the real harms are, where the lines need to be drawn.
I’m not holding my breath.
Beyond The Algorithm
Sex work has never been about “man pays to see hot woman naked.” That’s the story people tell when they want to reduce this work to bodies and transactions, when they want to pretend there’s no skill or labor or strategy or emotional intelligence involved.
The reality? You can find infinite free content online. The women who make real money in this industry, the ones who’ve been doing this for decades, who have client retention and genuine financial security, aren’t selling bodies. We’re selling the space where desire becomes real: where someone’s most private fantasies meet actual human witness.
AI can generate content. It cannot generate that.
What AI can do is flood the market with cheap substitutes that serve people who were never really our clients anyway. The ones who wanted objectified beauty without the complication of another person’s humanity. I say fine. I actually say GREAT! Let them have their algorithmic gooner content. Let the market segment itself.
Because here’s what’s going to happen: AI will eliminate the bottom tier of this industry; the tier based on looks alone, extremity and overconsumption (mostly by those who don’t even want to pay for it). It will create a space for parasocial relationships with chatbots for people who find actual humans too complicated.
And it will leave a premium market for people who want the real thing. Who understand that the value isn’t in the pixels or the script, but in the presence of another person who chooses to engage with your desire, who has boundaries and judgment and genuine reactions, who might surprise you or disappoint you or understand you in ways that feel dangerous and necessary.
That market? We’ll be fine. But how big it is remains to be seen.
The women who’ve been authentic, who’ve been good at communication and storytelling, who understand human psychology and power dynamics—well we’ve been competing with the free porn that’s existed almost since the internet began. And we won that competition by offering something else entirely.
AI just makes the distinction clearer.
So when people ask if AI will replace sex workers, I think: replace us at what? If you think this work is about having a body men find attractive, and being able to churn out content as fast as the blood can flow down between their legs, then sure, you’ll be replaced by a render that’s more perfect than any human could be.
But if you understand this work the way I do—as a practice of holding space for people’s most vulnerable fantasies, of creating safety within transgression, of offering genuine connection amid the performance—then AI isn’t your competitor.
It’s just another way of delivering free content to people who were never going to pay for the real thing anyway.
The industry will split. Let it. The creators who survive will be the ones who were doing this work right all along.
If you want to understand what people really want when they say they want something, or why authenticity matters more than aesthetics, or what happens when technology tries to replace human judgment—subscribe. This is just getting started.
Park, B. Y., et al. (2016). “Is Internet Pornography Causing Sexual Dysfunctions? A Review with Clinical Reports.” Behavioral Sciences, 6(3), 17.
Kantrowitz, A. [@Kantrowitz]. (2025, January 16). Sam Altman on AI companions: “There are more people that want [a deep connection with an AI] at the current level of model capability than I thought” [Tweet]. X.
Skjuve, M., et al. (2021). "My Chatbot Companion - a Study of Human-Chatbot Relationships." International Journal of Human-Computer Studies, 149, 102601.