I pretend to be OnlyFans models online

0:00This BBC podcast is supported by ads outside the UK.

0:30customers get 15 % off their first order. That's K-A-C-H-A-V-A dot com code NEWS. Craving the coffee flavor you love, but without the caffeine? Cachava's got you covered with their newest coffee flavor. This all-in-one nutrition shake delivers bold, authentic flavor crafted from premium decaffeinated Brazilian beans. Quality nutrition shouldn't be complicated. Just two scoops of Cachava's all-in-one nutrition shake, and you've got 25 grams of protein, 6 grams of fiber, greens, and so much more. Whether you're craving that coffee taste to kickstart your morning ritual or as a nutrient-packed reward to round out your afternoon, cachava keeps you fueled and satisfied wherever your day takes you.

1:10Plus, it actually tastes delicious. No fillers, no nonsense, just the good stuff your body craves. And for the times you feel like switching it up, you've got seven flavors to choose from, all with the highest quality ingredients. Treat yourself to the flavour and nutrition your body craves. Go to Kachava.com and use code NEWS. New customers get 15 % off their first order. That's K-A-C-H-A-V-A.com, code NEWS.

1:38Shiona McCallum:Welcome to Tech Life on the BBC World Service, the programme about how tech affects all of us every day. I'm Shona McCallum and today we'll be talking about OnlyFans, the huge online platform But are subscribers really chatting to the creators they subscribe to or someone else? So we ask their name, their age, are they comfortable? And then we ask what's their work. So basically we ask what's their work because that's when we could see how much we could make sales out of them. One of the ways AI companies say they will change the world is in speeding up science research. But can you stop AI slop muddying the waters?

2:21The best science is still going to be done by scientists who have devoted their lives to their field of study. They're going to be the best at guiding the AI. We see this day to day at understanding its output and really importantly at verifying its output.

2:34Shiona McCallum:And the US government has been criticised for mixing real footage of strikes in Iran with clips from video games. We take a look at what this says about the way war is sold to voters. It's actually a huge mistake to think that simply because games themselves aren't real, that means they don't have real world effects. They absolutely do.

3:15Shiona McCallum:Just a warning if you're listening with children, our top story contains adult themes. Often on Tech Life, we cover the hidden human work behind many online platforms, such as the people moderating harmful content or manually training AI models. Today, we're talking about OnlyFans, a platform where people can pay to subscribe to content creators, which has become most widely known for its pornography. One of the ways performers can make money on OnlyFans is to have private chats with subscribers who pay to receive sexually explicit photos or videos of the model. But sometimes those online chats are outsourced to other people pretending to be the performer.

4:00Shiona McCallum:Tech Life's Chris Vallance has been talking to one of those so-called chatters and he joins me now. Chris, tell us about what you've been looking into. Well, you may have noticed, Shona, that the British tabloids are filled with the stories of the high earnings of the top OnlyFans stars. but not everyone involved in this world is making big money. Many models employ agencies who hire so-called chatters who pretend to be them in online messaging with fans, persuading them to buy images and video. Much of this work is done by people in the Philippines earning a few dollars an hour. I've been speaking to a woman there who's worked as a chatter.

4:38To protect her identity, we're going to give her a fake name, Maria. We've also revoiced her words. I asked her to tell me about her chats with fans. So we ask their name, their age, are they comfortable? And then we ask what's their work. So basically we ask what's their work because that's when we could see how much we could make sales out of them. For some agency, they do have targets per shift. So for example, for this model, you have to make$300 sale because I was handling several models. Maximum is four models in one shift. How does it make you feel doing this work? It must be quite challenging.

5:20Actually, it's odd because I was actually joking around before, like I was made for this job. It's too easy for me. Because I feel like I do enjoy talking to people. And it's this job that I get to know more people. It's kind of icky when you think about it because you'll have to do sexting a lot of times, like several times in an hour. After the sexting part, you'll get to know them more. And at the end of the day, it's kind of tiring because you know that you're just there to pretend as the model. But then you get a glimpse of what they really are in real life. They are really nice people. They are good people.

6:00It's just they are really lonely. So yeah, I think that's the saddest part for me there working as a chatter. Most of the men Maria speaks to are in the US or UK, and as you heard, she has quite a bit of sympathy for them. And on one level, she enjoys the interaction and feels guilty about the deception involved in her work. But there are also times she's uncomfortable with the content of the chats, specific requests she found disturbing and which we can't broadcast, and fan behaviour she finds troubling. There are actually several fans that can be really pushy. They have really weird kinks or fetishes.

6:38So for me, that's a really challenging thing, working as a chatter, because you have to go along with it. You really have to deal with it. I mean, do you feel at the end of a shift or an end of period of work doing this, that it has any further impact on you? I think my tolerance is a bit high. But of course, there are days when I feel like, what the hell am I doing here? You know, because there are days that it would really take its salt on you, but there are days that you have more energy to push through. So it really depends. You just wish and hope that it's going to be bearable on that day.

7:19Shiona McCallum:And Chris, how about her own pay? How does that compare with the hundreds of dollars per day she must be making for the models she's pretending to be? Well, Maria told me she initially took the job to support her family during a period of low income. But her first job paid less than$2 an hour. A subsequent stint as a chatter paid not much over$3 an hour. That's less, a lot less, than the$300 in sales she made for the top earning models in one shift, which are passed on through the agency. Is there any sense in which you feel a bit exploited? Because you see how much money they're handing over to the model, and here you are getting, and then the lowest paid, not even$2 an hour.

8:03Yeah, knowing that they are making much money out of us. It's really kind of heartbreaking, especially because knowing that the agency is getting way more. And I'm the one that needs to work with those subscribers, with those fans. And how important was that money to you when you took that first job at that low rate? So I took that jump, even if it's really, it's really something that you wouldn't be proud of, you know. I mean, I keep on joking around that I was made for it. But looking back, I know that if I only had the means to say no, I won't do it. Well, some experts say this kind of work is in a legal grey area in the Philippines, which has strict pornography laws.

8:49In fact, Maria described the care chatters took over not revealing their own identity. The BIEN union, which represents workers in the outsourcing sector, told me there are some advantages to this kind of work as it allows for remote working and may pay more than some local entry-level jobs. But they worry about the unregulated nature of the work, meaning workers may face legal uncertainties, psychological risks and limited access to support or safeguards in their work environment.

9:20Shiona McCallum:And what about OnlyFans? Well, to be clear, the chatters don't work for OnlyFans. They work for agencies who are hired by the models. Only fans wouldn't speak on the record, but their terms of service say their contracts are only with content creators. The terms also say they don't endorse any third party platforms and that all legal responsibility for uploaded content rests with the creators. And we shouldn't forget about the fans. Some feel chatting is deceptive and have attempted to pursue legal action against only fans and the agencies who employ chatters. Thanks, Chris. Thank you. Craving the coffee flavor you love, but without the caffeine?

10:02Cachava's got you covered with their newest coffee flavor. This all-in-one nutrition shake delivers bold, authentic flavor crafted from premium decaffeinated Brazilian beans. Quality nutrition shouldn't be complicated. Just two scoops of Cachava's all-in-one nutrition shake and you've got 25 grams of protein, 6 grams of fiber, greens, and so much more. Whether you're craving that coffee taste to kickstart your morning ritual, or as a nutrient-packed reward to round out your afternoon. Cachava keeps you fueled and satisfied wherever your day takes you. Plus, it actually tastes delicious. No fillers, no nonsense.

10:38Just the good stuff your body craves. And for the times you feel like switching it up, you've got seven flavors to choose from, all with the highest quality ingredients. Treat yourself to the flavor and nutrition your body craves. Go to cachava.com and use code NEWS. New customers get 15 % off their first order. That's K-A-C-H-A-V-A dot com, code NEWS. Craving the coffee flavor you love, but without the caffeine? Cachava's got you covered with their newest coffee flavor. This all-in-one nutrition shake delivers bold, authentic flavor, crafted from premium decaffeinated Brazilian beans. With 25 grams of protein, 6 grams of fiber, greens, and so much more.

11:16Treat yourself to the flavor and nutrition your body craves. Go to cachava.com and use code NEWS. New customers get 15 % off their first order. That's K-A-C-H-A-V-A dot com code news.

11:39Shiona McCallum:You're listening to Tech Life on the BBC World Service with me, Shona McCallum. We've been talking about driverless cars for the past few weeks with lots of messages from you. This one came in on WhatsApp from Zev Farkas in Jerusalem. Zev heard last week's edition where we read out a message from a listener who said driverless cars are not safe at any speed. Zev disagrees and says autonomous vehicles take the most dangerous part out of the car, namely the person behind the wheel. And David Walker in the UK emailed us to say if two driverless cars are about to crash into each other, then they could have a digital conversation on how to avoid each other.

12:20Shiona McCallum:Nice thinking. Well, maybe they could at least cause the least damage. He suggests that you could pay more for your car to forecast potential injuries or even call the emergency services. Well, thanks to everyone who gets in touch with us. If you want to carry on the conversation on driverless cars or anything else that we talk about here on Tech Life, send us a WhatsApp. The number is plus 44 330 1230 320. Or you can email us. Our address is techlife at bbc.co.uk.

12:56Shiona McCallum:Scientists have always relied on powerful tools to push the boundaries of discovery, from microscopes to supercomputers. Now artificial intelligence is becoming the latest addition to that toolkit. Researchers are using it to speed up experiments, analyse data and even help solve problems. But critics are warning about a flood of so-called AI slop, unreliable or fabricated material. So does this new tech accelerate discovery or risk muddying the waters of research? I'm Kevin Wheel, VP of Science at OpenAI. My team and I focus on accelerating science with AI. Yeah, I'm Christian Köster. I'm Associate Professor in the Department of Computer Science in Oxford, and I'm working on algorithm design and mathematical analysis of algorithms.

13:47Just in the last, like, say, six months since we launched GPT-5, we've started seeing examples of AI actually answering some of the hardest scientific questions, mathematical questions, and beginning to go beyond the frontier of human knowledge and answer even open questions. You know, you think about what a scientific workflow would be. You have ideation, you're thinking through things, you're trying to solve a problem. In some fields, you may then have experimentation where you're actually in a lab trying to create something. And then you're writing up your results and you're collaborating with other scientists and sharing the output of your work.

14:26AI can help at every one of those phases. You suddenly have this incredibly knowledgeable, infinitely patient, always-on partner to help you think and debate. A scientist I was talking to recently called AI a metal detector for hypotheses, which I just loved.

14:42Shiona McCallum:And are you using AI in your work? Yeah, I am using GPT-5 to help me with my research. I mean, on the one hand, it's making me faster and making progress on them. And on the other hand, I think there are also some questions that I might not be able to solve just on my own. But also, I would not be able to solve it on its own. But by combining the skills of AI and myself, I can make progress on those questions. And Kevin, I'll come to you again, because there are concerns about AI slop, You know, that low quality, automated, AI generated content that clearly you don't want in science. How are open AI combating that?

15:28I think a lot of this points to the value of scientists and mathematicians and expertise. Because even with AI, you just heard Christian say it, it's not about AI or about scientists. It's about AI and scientists together. The best science is still going to be done by scientists who have devoted their lives to their field of study. They're going to be the best at guiding the AI. We see this day to day at understanding its output and really importantly at verifying its output. Things like peer review are still critical.

15:59Shiona McCallum:I guess some scientific fields are already struggling with this kind of flood of AI generated papers. We know that it can make up figures, citations. I mean, I just wonder, is there a risk that this AI slop could dilute the scientific discourse? None of this diminishes the value of peer review, right? You still want peer review. It's still critical that people are looking. If you're going to cite something, you better make sure that it's real. Whether you use AI in your work or not, you as the scientist are responsible and accountable for the output. I think Christian would agree. when you look at some of these conferences that are getting a huge amount of influx of papers, some of which are low quality generated by people who don't have the background.

16:46Again, I think one of the tools that we should be using to help discriminate between low quality work and great work is actually AI, because you can employ GPT-5 and very quickly separate a lot of the low quality work from the great work.

17:02Shiona McCallum:And Christian, we know that peer review and searching through literature, grant evaluation, you know, looking through all these different areas of science can be quite time consuming. I just wonder with the kind of addition of now AI generated content, it could become more tedious because you got more to, you know, more research to kind of sift through. Have you got a growing volume of AI-generated submissions? Have you noticed that as being an issue? So far, I haven't noticed it yet. I mean, it is true that the number of submissions are going up, but conferences are also enforcing some stricter policies, such as rejecting immediately a paper where it turns out that they have made up citations in the reference list.

17:50Yeah, there are also some requirements now of disclosing AI in some conferences.

17:56Shiona McCallum:And how do you ensure that you're using it as a tool then to help your work as opposed to replacing the expertise? I go to the AI system and I ask it about the research that I'm doing in a similar way to how I would talk to a collaborator. But the way that it responds to me is often different from how a human collaborator would respond because it has different skills. skills. I mean, on the one hand, it has this much broader knowledge where it can point me to other research that I don't know about, but it doesn't have the deep expertise that I have in some parts of my work. So for those parts, I'm still needed as a human to complete the research project.

18:38Sometimes it can suggest, for example, an intermediate step in a mathematical proof that I'm trying to attempt, but then it's unable to actually execute this proof. And I have to still do a lot of thinking in order to get there.

18:51Shiona McCallum:I wonder, Kevin, if you could tell me about how you assess the possibilities that models could perhaps accelerate that research into a dangerous domain or enable perhaps a dangerous discovery. So what are you doing to mitigate that? Safety is foundational to everything that we do. And we have model safeguards around biological research and other areas of sort of dual use research that apply to all of the models that we create. Our models undergo extensive internal and external testing. We're working with, you know, everyone from external red teamers. We work with the UK AI Safety Institute. We publish findings.

19:32We try to create industry standards. So, I mean, this matters a lot to us. And we put a huge amount of work into making sure that every use of our models stays safe, that you can't use them to take research in the wrong direction.

19:49Shiona McCallum:And do you think that there needs to be more external standards governing the use of AI in science? That's a good question. We're very open to those kinds of discussions. We work with a lot of external bodies already, you know, government bodies as well as scientific bodies and like I said external red teamers. So try and get as much input as we can from outside of our walls even as we put a lot of resources from inside our walls into making these models safe. And obviously there's been a lot of high profile people even within AI companies telling us about their concern the way that the industry is going.

20:32Shiona McCallum:I just wondered Kevin what's your thoughts on those voices and how do you address those concerns that AI could be a force for bad? I have never worked at a company that cares as much and has put as much resources into keeping our products safe and making sure that their use is beneficial as OpenAI does. Now, are we going to get it right in every case? This is new technology. We're learning as well, but I can tell you there is so much focus on making sure that our models are safe. Thanks to Kevin Wheel and Christian Coaster. Just to mention, this was recorded before OpenAI signed a deal with the US government to use their technology in the military and intelligence services.

21:26Shiona McCallum:Communicating about conflicts is one of the main jobs for world leaders. One of the strategies from the current US administration has been to post images and videos using lots of references to Hollywood films and video games. Last week, it cut together real footage of strikes in Iran with clips of gameplay from the game Call of Duty.

21:58Shiona McCallum:The White House has taken from other parts of gaming culture too including from Halo and Pokemon That led the Pokemon company to issue a statement saying it was not involved in creating any of these memes and did not give permission to use its intellectual property This week, Halo voice actor Steve Downs posted on X demanding the White House remove a clip which uses his voice in one of its videos, calling it disgusting and juvenile war porn. At the time of recording, Microsoft, who owns the studios behind Call of Duty and Halo, has not responded to our requests for comment. By using gaming memes and clips, what is the White House trying to say about its approach to the conflict in the Middle East?

22:41Shiona McCallum:Dr. Aggie Hurst is an expert in international relations and author of the book Politics of Play, Wargaming with the US Military. What's happening here is that the Trump administration is seeking to curate and control narratives about their policies, particularly when it comes to foreign policy and also immigration. As you say, the recent video we're discussing today focuses on Iran, but similar artefacts have been created around things like the ICE immigration enforcement that is, of course, so controversial in the States at the moment. Do you think it's tasteless given what's at stake here, given human lives could be lost and have been lost?

23:20I think it's absolutely right to be outraged at the use of games in ostensibly real-world operations footage. But what I would add to the discussion is that leaning on the justification that war is not a game actually obscures why the Trump administration is using games and gamers for these political ends. Now, we usually think of games as not real. They're just for fun. There's no real world consequences. You know, if you kill somebody in a game or if you invade a country, nobody gets hurt. We have that familiar trope. It's only a game. But I would say that it's actually a huge mistake to think that simply because games themselves aren't real, that means they don't have real world effects.