From GPUs To Guilt: AI porn is the elephant in the server room

Sam Altman has decided that five years is enough of a head start for the porn companies.

If you missed it, last week, OpenAI’s CEO Sam Altman broke this news on X:

“In December, as we roll out age-gating more fully and as part of our ‘treat adult users like adults’ principle, we will allow even more, like erotica for verified adults.”

What happens when code learns your desire better than you do—and never says no? We dive into the explosive rise of AI-generated pornography and deepfakes, exposing how ultra-realistic content collides with consent, identity, and the fragile social skills that make intimacy possible. We ground the conversation in research showing that the vast majority of deepfakes are nonconsensual porn, then map out the real human toll: shame, humiliation, and a loss of control that lingers long after a clip goes viral.

From there, we unpack the engine behind the explosion. Generative models don’t just recommend content; they manufacture it to your exact preferences, creating a “perfectly obedient” loop that pushes novelty and extremity. We explore why this personalization can rewire arousal patterns and make real relationships feel slow or difficult by comparison. We also examine the darkest frontier—AI-generated CSAM—where the “it’s just synthetic” defense collapses both ethically and legally, and platforms face a reckoning for safety failures at scale.

We trace the adult industry’s head start with AI and the market pressure pushing mainstream companies to relax erotica policies. Alongside that, we highlight emerging solutions: hard-coded ethics that refuse nonconsensual or abusive roleplay, provenance and face-rights controls, and friction around sensitive prompts. Finally, we walk through proposed laws focused on speed and accountability—fast takedowns, civil penalties for creators, and state-level rights to sue—while asking the bigger question: if instant, customized fantasy becomes the baseline, what happens to consent, negotiation, and care in real life?

If this conversation resonates, follow the show, share it with a friend, and leave a review with your take on the strongest guardrail we should build first.

Leave your thoughts in the comments and subscribe for more tech updates and reviews.

Transcript

Ida 0:05

Welcome back to the deep dive. For years, AI was sort of the helpful office intern, right?

Setting The Stakes

Allen 0:11

Right. Writing your boring emails, summarizing meetings you zoned out of.

Ida 0:15

Exactly. But now it feels like AI has left the office and it's uh walking right into the most private corners of human desire, and things are getting very complicated very fast. So today, we're diving deep into the explosive, often whispered about world of

Defining Deepfake Porn

Ida 0:31

AI-generated pornography and deep fakes. The whole thing feels well, it feels both bizarrely inevitable and like total digital anarchy at the same time. It's creating a real crisis of consent, of identity, and yet we're all still struggling to just talk about it honestly.

Allen 0:48

Yeah. There's a huge taboo.

Ida 0:49

A huge one. I mean, you know what I'm talking about. AI porn is the elephant in the server room.

Allen 0:53

That's the perfect way to put it.

Ida 0:54

Everyone in tech knows it's what's really driving the demand for GPUs for faster models, but we all just pretend it's some you know fringe concern, all while ignoring the ethical wreckage it's causing.

Allen 1:04

And that collective head-in-the-sand attitude is exactly why this deep dive is so necessary. Our mission today is to just pull back the curtain.

Ida 1:13

Right.

Allen 1:13

To hold a mirror up to this technology and analyze the sources you've shared with us. We need to unpack this tension between, say, hyper-personalized fantasy on one hand and the total collapse of consent on the other.

Ida 1:25

Aaron Ross Powell And we're pulling from everything legislative proposals, ethical reviews. Trevor Burrus, Jr.

Allen 1:30

All the way to raw visceral posts from Reddit addiction forums. We need the whole picture.

Ida 1:34

Aaron Powell Okay, let's get into it. And let's start with uh the basics. For anyone who still thinks this is just like goofy face swaps on memes. Explain what deepfake porn actually is.

Allen 1:45

Aaron Powell Sure. At its core, you're using advanced algorithms, they're called jans, to superimpose one person's face onto another person's body in a video or photo.

Ida 1:53

Aaron Powell But the key is the realism, right? It's not like it used to be.

Allen 1:57

Not at all. That's the difference now. The algorithms have become so sophisticated that the final product is often seamless. You genuinely can't tell it's fake.

Ida 2:05

Aaron Powell So it's a perfect illusion.

Allen 2:07

Aaron Ross Powell A perfect illusion of reality. Used to show someone doing something sexual that they uh never did and certainly never consented to.

Ida 2:15

Aaron Ross Powell And this brings us to the statistic that it defines the whole problem. It shows this isn't a side effect of the tech. What percentage are we talking about here?

Allen 2:23

Aaron Ross Powell The figure, and this is from researchers at sensity.ai, it's

The Consent Catastrophe

Allen 2:27

just staggering.

Ida 2:29

Okay.

Allen 2:29

It's consistently between 90 and 95 percent.

Ida 2:31

Aaron Powell 95 percent.

Allen 2:32

90 to 95 percent of deep fake tech is used specifically to create non-consensual pornography NCP.

Ida 2:40

Aaron Powell So this technology, for all its potential, has a fundamentally malicious default setting.

Allen 2:46

It does. When it's left unregulated, its primary function is abuse.

Ida 2:50

That number just makes it so clear. This isn't a bug, it's the feature. So if it's so powerful and its main use is so clear, why are the big AI companies so far behind on filtering this stuff? Is it a technical problem or do they just not care?

Allen 3:03

It's a bit of both, but I'd say it's mainly a prioritization problem.

Ida 3:06

How so?

Allen 3:07

Well, a lot of these models are open source. So bad actors can just take them, fork them, and modify them specifically to get around any safety features.

Ida 3:15

Uh, okay.

Allen 3:16

So a big company might spend a fortune trying to stop their own model from making this content, but they can't stop all the modified versions out there that are designed to look perfectly real. The game of whack-a-mole.

Ida 3:26

And the real world impact is just devastating. I mean, it moves from a statistic to a personal catastrophe. Our sources lay out how it's being weaponized, yeah. Like the classic revenge porn scenario, but supercharged.

Addiction And Hyper‑Personalization

Ida 3:39

A rejected classmate takes a few photos from social media and creates a hyper-realistic porn video.

Allen 3:45

It completely changes the risk of just having a public profile. And the pool of targets is basically infinite.

Ida 3:53

You're talking about that viral thread from the OnlyFans model, Laura Lux.

Allen 3:57

Yes, that was chilling. She talked about this avalanche of men saying they won't need sex workers anymore because they can just AI generate a video of, and this is a quote the hot girl that serves them their coffee every day.

Ida 4:09

That quote. It shows how AI is being used to just eliminate the need for consent from the equation of desire entirely. Any picture of you becomes a resource.

Allen 4:18

A resource for someone else's fantasy. Exactly.

Ida 4:20

And the trauma is so clearly real, but you still see this weird philosophical argument crop up.

Allen 4:26

Oh, I know the one.

Ida 4:27

People saying, well, no one's physical body was hurt, so is the harm really that bad?

Allen 4:31

It's such a fundamental misunderstanding of trauma and identity. The harm is absolutely, undeniably real. One of our sources was a woman who saw her own smiling photos, pictures she loved, edited onto graphic images of gang rape and strangulation. She described the feeling as the shock of seeing yourself be violated.

Ida 4:54

It's a violation of your sovereignty.

Allen 4:56

Absolutely. The shame, the humiliation, the total loss of control. It leads to very real-world consequences, depression, suicidal thoughts. The psychological wound is incredibly deep.

Ida 5:06

To really get why this is exploding, though, you have to look beyond the trauma and at the engine driving it, the algorithm. Right. The internet used to optimize for clicks. Now with AI, it's optimizing for pure customized desire. And the psychology here gets really disturbing.

Allen 5:22

It really does. It's the ultimate feedback group. The power is in the hyper-personalization.

Ida 5:27

What do you mean by that?

Allen 5:28

Well, old school porn was static. You picked a category, you watched it, AI content learns you, it tracks your preferences, your specific fetishes, and it just renders them for you instantly.

Ida 5:39

So fantasy can be perfected.

Allen 5:41

Perfected. Updated every single night and tailored to hit your specific dopamine triggers. The sources call it a perfectly obedient experience.

Ida 5:49

And that perfection is what creates this unique, powerful form of addiction. We saw this all over the forums you researched. All over the Users are literally calling AI porn the devil himself. The most addictive form of porn because it can fulfill, and this is another quote, all your filia.

Allen 6:04

The specificity is the trap. In real life, desire requires negotiation, right? There are boundaries, compromise.

Ida 6:11

Yeah, other people are involved.

Allen 6:13

Exactly. AI just erases all of that. And that immediate, perfectly tailored fantasy becomes the new baseline. And this boundary removal, it's not just about consent, it extends to the physical acts themselves.

Ida 6:26

You pointed this out that AI lets creators simulate acts that, quote, could otherwise not be done due to the limitations of the human body and the risk of severe injury.

Allen 6:36

Yeah, that's boundary elimination at its most extreme.

Ida 6:39

It's terrifying.

Allen 6:40

It is. Because when the only limit is computing power, not human safety or ethics, the content will always, always drift toward the extreme. And that has to have implications for how users then view real-world relationships.

Ida 6:53

And we have to talk about the darkest, most illegal implication of all this. The creation of AI-generated child sexual abuse material. CSAM.

CSAM And Platform Failure

Allen 7:01

This is where that argument of it's just synthetic completely falls apart, legally and morally.

Ida 7:06

Right.

Allen 7:06

US federal law is very clear. CSAM is any visual depiction of sexually explicit conduct with a minor. Yeah. It doesn't matter if it's a photo or an AI image, creating it, sharing it, it's a major federal crime.

Ida 7:16

Because it still sexualizes children, the harm is real.

Allen 7:20

The harm is absolutely real, both to the public and to any child victims who might see their own likenesses used.

Ida 7:27

And this is a problem even for the biggest platforms. I mean, Lenza got a ton of criticism for this, right?

Allen 7:32

A ton. They had a no-nudes policy, but they couldn't stop users from generating CSAM.

Ida 7:37

Which really highlights the core problem.

Allen 7:39

It does. The platforms want the creative power of AI, but they don't want the responsibility of moderating the darkness it can create. They'll say it's about user intent, but the models are so powerful now that anyone with a phone can create this stuff, often just by lifting a kid's face from Instagram.

Ida 7:56

Leading to an unmanageable explosion of this material.

Allen 7:59

An exponential rise. It's a nightmare.

Ida 8:01

Okay, let's pivot a little and talk about the industry itself. Because historically the adult industry has always been on the cutting edge of tech.

Allen 8:08

Oh, absolutely.

Industry Head Start And Market Pressure

Allen 8:09

Streaming video, online credit card payments, VR.

Ida 8:13

They pioneered all of it. So how does that history give them a head start in the AI race?

Allen 8:18

It means that while a company like OpenAI was still busy perfecting its language models, all these decentralized, let's say, shady players were already years ahead on the porn front.

Ida 8:29

They had a three to five year head start.

Allen 8:31

At least. And they were operating with no ethical guardrails, no regulations. They built the tools and they proved there was a massive market for this custom content. Trevor Burrus, Jr.

Ida 8:40

Which basically forced the mainstream platforms into a corner. Trevor Burrus, Jr.

Allen 8:43

You put them in a totally reactive position, yeah? Scrambling to catch up.

Ida 8:47

Which brings us to that huge pivot from open AI, where they decided to start allowing erotica. They tried to frame it as, you know, treating adult users like

Hard‑Coded Ethics Vs. “Just A Tool”

Ida 8:55

adults.

Allen 8:56

Which nobody bought for a second.

Ida 8:58

No. Critics immediately said it was just a business move.

Allen 9:01

It was pure market pressure. One source put it perfectly. Something like open AI needs erotica more than Pornhub needs AI.

Ida 9:09

Huh, that's good.

Allen 9:10

It's true. They saw this massive economic engine of hyper-personalized desire and realized they were leaving too much on the table. They had to break their own taboo to keep pace with the unregulated market.

Ida 9:22

But inside the industry, we are seeing different ethical approaches pop up. Some developers are trying to build in real guardrails. Tell us about the hard-coded ethics idea.

Allen 9:33

It's a fascinating approach. It's about treating the AI as more than just a tool, almost like a digital partner that has to respect boundaries.

Ida 9:41

So how does it work?

Allen 9:42

We saw an example with an AI romance app, mypeeach.ai, which was founded by a former sex worker. They hard-coded ethics into their model. The AI is trained to refuse any role play that simulates abuse or non-consent.

Ida 9:57

So you can ask, but it will say no.

Allen 9:59

It'll say no, and it will explain why the request violates its ethical framework. It's an attempt to build the concept of digital consent right into the code.

Ida 10:07

Aaron Powell That is a really interesting idea, but then you have the complete opposite argument.

Allen 10:11

The counterpoint, yeah.

Ida 10:12

Which basically says the AI is just a tool, a complex one, but still a tool.

Allen 10:16

Correct. Other developers just say an AI is an unconscious thing. They literally equate it to a dildo.

Ida 10:21

A piece of plastic, or in this case, a bunch of coat.

Allen 10:24

Exactly. And their logic is since the AI can't consent and no real person is physically there, the whole concept of consent just doesn't apply.

Ida 10:32

Which, as we've discussed, completely ignores the fact that this tool is creating traumatic, perfect digital copies of real people without their permission.

Allen 10:41

Aaron Powell It's a massive convenient blind spot in their logic.

Ida 10:44

Aaron Powell Okay, so with all this chaos, regulators are playing a frantic game of catch up. Let's get into the legislative action, because this is where the response is finally starting to take

Laws Racing To Catch Up

Ida 10:53

shape.

Allen 10:54

Aaron Powell It is, and the focus is really on two things liability and speed.

Ida 10:57

And so on the federal level, you have the Take Eye Down Act.

Allen 11:01

Right. That act does two key things. It criminalizes knowingly publishing a non-consensual deep fake, and this is crucial, it forces platforms to remove the content within 48 hours of being notified by the victim.

Ida 11:14

That 48-hour window is huge. In the age of virality, two days can feel like a lifetime, but it's a start.

Allen 11:20

It's a massive operational challenge for platforms, but it's essential for trying to limit that initial catastrophic harm.

Ida 11:27

And then there's the proposed defiance act, which tries to deter the creators themselves.

Allen 11:31

Yeah, that one hits them in the wallet. Hard. It creates significant financial liability.

Ida 11:36

How significant.

Allen 11:37

The act would let victims sue the creators for up to $250,000, especially if it's tied to stalking or harassment.

Ida 11:45

A quarter of a million dollars, that's a real deterrent.

Allen 11:48

It's a game changer. It attaches a very tangible, very painful consequence to an act that's been historically hard to prosecute.

Ida 11:56

And the states are moving fast, too.

Allen 11:57

Oh, absolutely. California passed legislation criminalizing the creation of AI porn without consent. And New York's Hynchy Law is even more specific. It makes creating or sharing these deepfakes a crime. And it explicitly gives victims the right to sue.

Ida 12:13

So the law is finally starting to recognize the unique nature of this digital violation.

Allen 12:18

It's trying to close the gap that technology created overnight. So to sort of tie all this together for you for everyone listening, what we're doing is designing

Autonomy Trap And Final Challenge

Allen 12:25

a world where our fantasies can be infinitely customized, instantly available.

Ida 12:30

But there's a cost.

Allen 12:31

There's a massive invisible cost. And that cost is the disintegration of our digital identity, of our personal autonomy. The sources call it an autonomy trap. The second you put a photo of yourself online, that image becomes a perpetually malleable resource.

Ida 12:48

Aaron Powell A resource that anyone with a laptop can manipulate without your knowledge.

Allen 12:52

Without your knowledge or your consent.

Ida 12:53

That phrase, perpetually malleable resource, is honestly chilling. It basically says the only way to have digital privacy is to not exist digitally.

Allen 13:01

Pretty much.

Ida 13:01

So the ultimate question for you to think about isn't just how we regulate the stuff or if a 48-hour takedown is fast enough. The deeper challenge is this. By making customized, simulated pleasure the priority, are we eroding the very social skills we need for real world consensual intimacy? What happens to actual human relationships when the baseline expectation becomes an instant, obedient, and perfectly synthesized fantasy? That's the real challenge AI porn forces all of us to confront.