Inside PornHub's Safeguard Strategies: Exclusive with ECP Partners Soloman Friedman & Sarah Bain

Kim, you really asked the right question, because you asked about procedure and technology, and it's really important to emphasize that even though we are a tech company and we have some incredible tech solutions — both those that we've developed ourselves and those that we use from third-party providers — technology is not enough in this. What we've actually done is we've made a massive investment in people. So let me frame things: who we are, what our company looks like. Aylo is a company of roughly 1,400 employees, which you have to understand is completely out of scale for the adult industry. That's a serious undertaking, and the reason it is so out of scale is because of how seriously we take trust and safety. Of those 1,400 employees, 20% work full-time in trust and safety. If you asked Twitter, Facebook, what are your trust and safety numbers like, it's minuscule compared to that in terms of proportion. That's a massive investment. This cannot be done with technology alone. I'm going to talk about how it's powered by people — whether they're frontline moderators, identity verifiers, or all the way up to those developing policy, managing, and directing. We put our money where our mouth is. It's not cheap to run a trust and safety program that I'm about to describe, but it is critical. As the ownership group, we never for a moment would have contemplated buying this company if what I'm describing didn't exist. It just would have had no interest to us, would not have been a platform we could invest in. Let's talk about what that looks like. As of today, there is no content on any of our content-sharing platforms that was not uploaded via the verified uploader program, sometimes also called the model program. That means that for every single uploader, we have verified their identity. We verify identity because by verifying someone's identity, you verify age, ID of course, but also consent. In order to become eligible to upload content, we're not YouTube — you can't just create an account and start uploading content. You have to apply, and the first stage of that application process is reviewing the terms of service, understanding what the site permits and does not permit, but then verifying your identity. We use a third-party provider called Yoti for the first step of that identity verification. They provide verification for governments around the world, for banks, insurance companies, car rental agencies, all sorts of major players. What we require is a biometric scan of the uploader's face in motion to ensure it's not a screenshot or a video of someone, then government ID — not a screenshot, has to be real — and then the technology matches both. We verify that you are a live person, a real person, we verify that you have government ID that matches the details you've submitted, and then that your face matches both of those. If you pass those checks, the application is then passed to one of our human moderators to confirm. I want to underscore that if you don't pass those checks, you don't move on. But if you pass that automated check, you then move on to a human moderator who is looking for other red flags. We're looking at everything from the authenticity of the document, whether they're holograms — we have all this technology to help match that — but fundamentally we're verifying identity. I want to pause here to say that a few months ago, AI-generated deepfakes of Taylor Swift erupted like wildfire all over Twitter and other social media platforms. Those would never have passed even the most basic cursory review on our platform, because of course we don't have Taylor Swift's biometric scan of her face and her ID. That would have been the end of anyone attempting to upload that. So something that people are rightfully concerned about — deepfakes are an extraordinary concern — but that could never happen to that level on our platform because we don't have the ID documents of the individual depicted. So that's verifying the identity of the uploader. Let's move forward. If all those checks are passed — we verify you're a real person, you match your ID, you're over 18, you've provided consent, etc. — you are now eligible to attempt to upload content. And I use that word 'attempt' very precisely, because unlike YouTube or Facebook or Twitter or Reddit, you cannot just upload to the platform. You can attempt to upload, which means you submit a piece of content for moderation. When you submit it for moderation, the following things happen. First, before it ever sees human eyes, that content is run through 10 different engines. We subscribe to every single content fingerprinting service made available to scan for child sex abuse material. We subscribe to engines called Google CSAI Match, Safer by Thorn, and NCMEC — the National Center for Missing and Exploited Children. They all have databases where they have scanned known instances of child sex abuse material, they've fingerprinted them, and then we can match all of our content against those fingerprints. If there's a match to any of those, that's the end of it. We make a report — I'll talk about how we report when we get to suspected child sex abuse material — but that's the end of it. It goes through about 10 of those engines. By the way, because we actually believe in transparency, we publish twice a year a transparency report in which we describe all of this technology. I know we're telling your listeners this and many of them are hearing it for the first time, but we have been saying it publicly, as clearly and explicitly as we can. Someone once asked me at a conference, 'Wow, in your transparency report you publish all the tools that you use — aren't you worried that someone's just going to copy what you do?' I say no, that's what we're hoping for. I'm not worried about it. I want every adult and social media company to copy the processes that we use. That's why we don't hide them — we are very transparent about them and publish them openly, including the numbers of how many uploads are stopped at what stage and reasons for removal of content. Now, if the content passes all of those automated checks, then and only then is it provided to a human moderator for review. We have a team of human moderators that work in shifts, 24 hours a day, 7 days a week, in person. They work in a secure facility because of course they are charged with reviewing explicit content and reviewing people's personal identifying information, and they review it for moderation. What are they looking for? Number one, we are the only free content-sharing platform on the internet that at present requires all individuals featured in content to provide the platform with their identification and consent. That doesn't exist anywhere else. If there's five people in the video, the person uploading the content is verified, but they also have to provide verification for those five individuals in the piece of content they're uploading. Correct, every single time they upload, every single piece of content. Now, of course, individuals might not want to give their identifying information to the uploader — maybe they just have a business relationship — that's totally fine. They can submit it separately through the portal. But the moderator then matches the ID to the individuals depicted.

Yeah, great question, Kim. We treat content that doesn't show the face to the public the same as if it does show the face. In other words, the moderator has to be satisfied that it's the same person. Now to the public, they don't see the face, so they might think this person's face was never depicted, so the moderator never saw it. In fact — and I'll be happy to show you this when I show you our tour of trust and safety — moderators have all sorts of ways to confirm it. For example, they can request that the uploader submit a longer clip, not for publication, just for moderation, that does show the faces, even though that's not what's going to be shown publicly — clip it out, but the moderator will review that. If a longer piece of content isn't available, they have other things they can do. For example, if they see the individual depicted has certain distinguishing tattoos or marks, they can then request a full body and face photo of that individual so they can match the face to the body and then to the ID that was provided. So faceless content to the public might be faceless, but it's not faceless to us on the moderation side. And once again, that's why this dialogue is so important, because there's so much happening behind the scenes, including the fact that all of our moderators are full-time employees of the company. I know some of your listeners may have heard how other social media companies are outsourcing to call centers around the world. We do absolutely zero outsourcing of moderation. The kind of information we're dealing with is just too private and sensitive to allow for that. Our moderators — it's not sign up today and you're a moderator tomorrow. There's actually a three-month course before they start any actual work, and at each level we test them on content. Because they're doing a lot — they're looking at obviously age, ID, and consent, but those are the basics. For us, our terms of service and our moderation are not bound by what's legal and illegal. I always like to say, complying with the law is like putting your pants on in the morning — you don't get a gold medal for doing that. It's what's expected of you. But our moderators are looking for so much more than that. Let me give you some examples of what is banned and a moderator would be looking for. Obviously you have to be over 18 to be on the platform, but you can't pretend to be young. We treat a depiction of an underage person as if it's underage — whether it's roleplay, totally consensual — we don't allow it. We will not create a market for that kind of material. Sometimes people are more subtle — they don't just get up and say 'hi, I'm 17,' but they're acting in a way or they're in a bedroom that has art on the wall that looks like it's a young person, or there's contextual cues that say this is a young person. So a moderator's job is twofold: to look at content, but also to look at context. But we take it even further than depictions of underage. We don't allow any depictions of either non-consensual activity or non-consensual recording. For example, even people pretending that it's non-consensual — absolutely banned. And by the way, we get uploaders saying 'this is our kink, this is what we're into, it's totally consensual.' We say, I'm sorry, we're not the platform for you. At the same time, we take the exact same approach to depictions of non-consensual recording — if it pretends to be a spy cam or a voyeur scenario, a hidden recording, we don't allow that. We don't allow even pretend depictions of extortion or intimidation, etc. Similarly, we don't allow any depictions where a person cannot at all times withdraw consent. This happens in the BDSM community — perfectly legitimate, perfectly legal fetish — but if a person is bound and they can't show non-consent, or gagged to the point where they can't show non-consent, we don't allow that. We don't allow depictions of intoxication by drugs or alcohol. Now this one's going to blow your mind — I never knew about this, I came from outside this world — we don't allow depictions of mind control or hypnosis, even if it's totally pretend. We don't allow it because what it is showing is a depiction of a sexuality that we're simply not the platform for. It's totally legal — in fact, there are Hollywood movies made all the time where these things are depicted in fiction and they win major awards — but we don't show that. Moderators are looking for many, many things. We don't allow depictions of serious violence that appear likely to cause serious injury or death, even though it's total acting. The moderators are doing a very comprehensive job, and that's one of the reasons, by the way, when a verified uploader uploads a piece of content to Pornhub, it can take hours to days to sometimes weeks to go through the moderation process. You're not going to have it up instantly. We simply cannot do that to maintain the standard of moderation that we require.