Name: Fake Experts, Real Risk: How Journalists Can Vet Sources in the Age of AI
Uploaded: 2026-02-18T17:16:14.781Z
Duration: 44 min 12 s
Description: Fake Experts, Real Risk: How Journalists Can Vet Sources in the Age of AI

Transcript for "Fake Experts, Real Risk: How Journalists Can Vet Sources in the Age of AI": K. Are we Hello. Just? wanna make sure. Hello. Hello, Shelby. Hello, Rob. I, and hello, everyone who's tuning in for today's discussion. I am Dan Simon. I'm the CEO and cofounder of Quoted, which is a platform that serves journalists and people working in the media. And I'm super excited to be joined today, firstly, by my colleague, Shelby Bridges, who's the head of user success at Quoted. And the reason that Shelby's on the call will be abundantly clear very soon and also by Rob Waugh from Press Gazette. And and, again, Good frontline experience is going to be invaluable, I think, for today. So I'm super excited to be joined by Shelby and Rob. Thank you both for joining me. Yes. Good to be here. So looking forward to it. Great. And so today, we're going to be talking about an issue that I think we see every day inside quoted, and we see talking to our media users. And Rob and and his colleagues at Press Gazette have been particularly dogged and focused on, which is fakery, AI, and authenticity or inauthenticity that is that is sort of happening, especially supercharged with with AI, the fake expertise, fake experts, fake brands. How is that impacting journalism? How are journalists dealing with it? What is the you know, what how is it changing their day to day life? And then and what what can we do about it? What can we do about it as an industry, and and what can we do about it as a society? What are the impacts? And so with that, I'm going to stop speaking. I'm gonna pass over and start and I'll start asking some questions of Shelby Rob, and I'm gonna begin by asking you both to tell us a little bit about your background, tell us about kind of your your career, And then and then in doing so, tell us a little bit about how you first came to encounter the problem of of of of fakes or or fake expertise. And, Rob, I'll I'll start with you. Well, yeah. Thanks. My background's, I'm a kind of freelance journalist. I've been freelance since the sort of age of the dinosaurs, and I worked for a lot of newspapers like the Daily Mail, Telegraph, and wrote about sort of technological subjects for Press Gazette and wrote for a whole variety of other clients, mostly in the technology space. I've been interested in AI long before it was fashionable. And my my experience of if they keep people, it actually it began maybe three years ago. I was doing some working for a marketing agency in Barcelona and we got offered a guest blogger and I just looked at her and I just was like, this person doesn't exist. And then a year later, I was writing for The Telegraph, is a newspaper in The UK, and I had to find a psychiatrist to talk about the psychological impact of identity fraud. And this psychiatrist answered almost immediately. And the thing was that I thought I I can't use the background she's given me because weirdly I mean, I thought even at the time, I thought this is odd. It's really odd. Her only employment that was listed was that she worked for a shop selling sex toys. And The Telegraph, if you're not from The UK, is quite a is quite a sort of proper respectable newspaper. And I thought, I cannot mention a sex shop in The Telegraph. The readers will have a heart attack, and they'll probably be killed by the editor. So I started looking at this person, and I realized, again, she does not exist. And I realized also at that point, despite the fact that she didn't exist, she'd been featured more than a 100 times, including in highly respectable British publications such as BBC and The Guardian and even American publications like New York Post. How about you, Shelby? Yeah. Tell tell us about tell I mean, that's obviously that's pretty shocking, Rob. And thanks. for bringing us up to date. And, Shelby, go go go ahead. So my background is sort of on the flip side of where Rob is seeing things. So I came from sort of a payroll, HR, SaaS platform technology back work background working in customer success. I think the the benefit of that is it gave me a lot of operational understanding of many companies and how they work on the back end, how they streamline workflows, processes, how they make their internal members of their staff available for content media type opportunities, and how often those top representatives within companies are actually not readily available to provide commentary. So then I started it quoted about four years ago, almost exactly. And it was a just moving into the customer success realm. This was the only thing I was focused on here. And that was maybe a bit broad. But after about six or seven months so I think that timeline aligns with you, Rob. It was like mid to late twenty twenty two was when I realized we'd found a specific person in our network. And I was like, this person is like a debt legal attorney, and they are commenting on everything and quoted. And not only are they commenting on it all, it's perfectly, perfectly answered every single time. Like, how can this person comment on things like city planning? And I think I have a slide to share in a little bit. I actually have the original pitch that I stumbled upon where I was like, this isn't right. And something's off. So that's where we started to realize we needed to have some checks and balances internally inside of Quoted and started realizing these indicators of what we needed to start looking for. And that has obviously scaled as ChatGPT, and everything has become much more accessible and has become even more intelligent. And I think that's initially, it started as you could have expertise in anything. But I think that's scaled now to you can be anyone, and you can create anyone, and you can create any brand and company. And that's where we kinda continue to see that scale up and get even more intelligent. Yeah. And and and we first question that we got, and I will formally hand over to q and a in about twenty minutes, but Jeffrey, in the comments said, you know, how are you defining fake? And you couldn't have teed up my next question better, Jeffrey, which is, you know, I did use a very broad term at the beginning of this, and I was like, fakes. There's lots of fakes. But but maybe, you know, I'd love to hear if if you either of you have a little bit more depth of experience with kinda what do we mean by fakes? What's the shape of the problem? Like, you know, what what are all these sort of are all of this is all of this problem equally sized or equally shaped, or does it come in sort of different flavors? I mean, I I. think I think there's there's a lot of depth to this problem. There's there's sort of out to nine fix where it's like AI generated profile image, copy mostly written by AI, have you know, there is no expert. There is no person. And, you know, those are the hardcore. And to give you an idea that the scale of it, certainly in Britain, we've uncovered well over a thousand articles featuring advice and advice tips and anecdotes from people who just fully do not exist. And then on some occasions, we've had the companies confess to us this person does not exist. In other occasions, you know, they've sent legal threats to us going, if you publish, we will sue, and then we published, and they didn't sue. So, you know, those are the hardcore. Then you go down to the next level where it's a purse there's there's it's a person, but they don't have the expertise that they claim. And we've seen quite a lot of ones. I mean, the ones we find most shocking are where it's like the company's SEO guy, but he's offering medical advice in a sort of, you know, as a a gut health expert or something like that and has no medic medical expertise. And then there's just sort of more nebulous ones. I mean, there's a lot where this advice is just run with experts at such and such company, And that has actually run-in newspapers, which as a veteran read old journalist, I find absolutely shocking that that has ever happened, but it does. And, I mean, what do you think, Shelby? Yeah. From our side, like, we're I I think the easiest thing for me, and maybe it's how my mind works, is we've kind of put placed them in little buckets, if you will. And so there's here, actually, I have some some resources. So I would say the most common thing that we see is it's a real expert. It's a real brand, but it's either fake representation and fake content that they're providing. And like I said, that was that's actually so this is that pitch that I initially said at the beginning, where it was a woman who's like a debt legal attorney. And she's commenting on this one about city planning for San Francisco. She's not from San Francisco, but she has this perfectly mapped out pitch here about how they could save San Francisco's downtown. And so it's very, very common because anybody could take a prompt, plug it into ChatGPT, and ChatGPT is going to make them look like the expert on that subject. So I'd say that's probably the most common, but I think maybe the easiest to see because that content sounds very shallow. There's a lacking of any type of quality in it. It sounds like a big echo chamber. It sounds like everybody else who's using the same way to submit that type of content. So this is obviously what we see on a regular basis. We commonly see this being individuals who hire, like, offshore assistance, offshore SEO help, lower lower rent type agencies that they're hiring to kind of get their foot in the door to start ranking their SEO as a business and not necessarily even knowing that what they're doing is potentially really hurting their reputation, but also really hurting their brand and the way that this person's putting them out into the world. So usually, when we talk to the actual experts in these scenarios, they have no idea the type of content they've been contributed to. And they have no idea that this person's kind of mass pitching them out to journalists all across the world on any type of topic. The second type of bucket that we have is a fake brand and a fake expert. So that then coincides to also be fake expertise as well that they're submitting. These ones, I mean, I guess for me, because I'm looking at it day in and day out, these also tend to be easy once you start to identify them. So these are I'm sure everybody has seen sites that are like bestplumbingcompany.com. Or it's often things that are very generic type labels or generic company labels. And they're things that they're trying to rank and build those domains so that they can turn around and sell them to other companies. So fake experts, typically fake personas. So if they're sharing a LinkedIn, anything like that, this is really where, you know, checking in at the credentials that are in those LinkedIn. Was that LinkedIn created yesterday? There's there's a lot of kind of barriers there. But those those websites tend to be very, very generic. It's not always quite clear on what products they're even selling or if they are selling a product, and they're just trying to get backlinks to it. But those those websites tend to always kind of just look like copy paste, change the colors, change what industry we're talking about today. And then the third one and one that I think is maybe the most interesting is fake experts, but real brands. And we're seeing this a lot. The easiest way to kind of think about it is like lead generation websites. So I'm sure you guys have all gone to a website, plugged your name in, and then all of a sudden, somebody from another company is reaching out knowing that you were kind of interested in that service. This example here is a company called Sparkly Made. They are a real company. You can look them up. They have locations all over The United States. They're a cleaning company. But they will go out and pitch as the cleaning expert with fake personas, fake individuals. Their whole website is actually made up of all fake individuals. But then their company itself is not the cleaning company. So they are not doing any cleaning. They are not cleaning experts. They are selling those leads and pushing them out to their partners in all of those different cities that they're located in. So this has been the easiest way for our team to kind of tactfully kind of break down what type of person are we dealing with. I would say maybe giving a little grace here. The fourth type of person would maybe be the person who's unknowingly submitting AI content and not realizing that they're not supposed to be. So they are in here repping themselves. They are doing that. They're trying to cut time. So we like to keep a portion of our business to assume best intentions and assume that we can just educate these people to help them understand that they should really be putting really clear, quality, candid content forward. Right. And and, I mean, we should say, obviously, there we give media on our platform the ability to check against AI, and we've heard directly that sometimes people don't mind a little AI. If it was in a bio, for example, you might expect someone to use GPT to tighten up a bio. But, obviously, when it comes to the actual expertise of that person, that's a different matter. Rob, I'd love to turn to you and ask we've sort of Shelby did a good job of explaining what the sort of the shape of the problem is. And as someone who's been out there on the front lines and is you know, you and Press Gazette have been leaders in kind of naming and shaming and trying to uncover this, what's the scale of the problem? Like, as a reporter, how often do you think your inbox or what percentage of your inbox today from just the from the from the universe, from the general population, how much of that is you know, what it what of that is is is real, and what of that is is fake? We did we did analysis with Neomam Studios. It's a content marketing company in The UK. They're campaigning against this stuff. And we did analysis with them of 250 stories chosen pretty much at random form from five non paywalled and kind of popular journalism sites. I I would use the word tabloid, but I used to work for a tabloid newspaper and I used to get wrapped on the knuckles by the editors if you ever said the t word. But they we found that one in 10 stories that are kind of an expert led stories. One in 10 have an expert that is completely fake. This is in a time period up to November. And we find that so one in 10, that is a completely nonexistent person or a an unnamed expert. So experts at such and such company, and they're not named. But we find that beyond that, there's a further I sort of 20 to 30% of those articles that we picked completely at random, which where the expertise is highly questionable. So it's often just an SEO person at that company writing the copy, or it's somebody who, for instance, you know, is at a sort of coupons company or, giving personal finance advice or, you know, advice that they are blatantly not qualified to give. And these these are being run uncritically in The UK and and in other places just by people who are looking for viral headlines. So these guys are supplying the the viral headlines that people want, which is, you know, I'm a carpet expert. Here's the here's the cheap trick you can use to make your house smell great, which clearly performed great on other things like Google Discover. And this rubbish is being run well, in the sample that we looked at, it's, you know, up to up to perhaps 40% of articles are pretty dubious, and 10% are the experts just completely fake. Yeah. I that's an that staggers me, to be honest, Rob. I thought you were gonna say one in 10 pitches you received were fake, not that your analysis had showed that one in 10 articles already being published contained fake experts. That seemed like a shockingly large amount of. misinformation and disinformation. It is. I mean, it and it's a failing on both sides because it's a fail it's you know, there are unscrupulous people who are using this method to try and just get SEO links at at any cost. And then on the other side, there's a fail failure of basic due diligence. I mean, these, you know, none of these stories should emerge. And I know I I could see in comments and people teeing up the q and a is that the logical questions we're gonna people are asking, okay. What do we do with it? What how do we tackle this? And I promise those who who are listening that that will be the sort of the third portion is we're gonna talk we're gonna talk about what what do we do to tackle it. But I just do wanna dwell a little bit more on the scale of the problem because, honestly, Rob, that that statistic is shocking to me, and I and I worry about you know, we we will talk a little bit about what kind of controls both human human inside a platform like Quoted, but I think, you know, the amount of bad actors that get kicked out and vetted and and flagged and so on, I gotta think that an average reporter's email inbox must be an absolute war zone, especially also today. I mean, also, given this volume, if you can if you can produce more content more quickly and leave it aside the whole fake thing, I would have thought that would increase the volume of noise that you receive in your inbox. Mean, that's certainly the case. I've spoken to journalists on big British publications, The Sun, The Times, others, and they've sent me photos of their inbox. And some of these agencies are sending, you know, 30 emails a day to each journalist. I mean, there's a guy in The UK who posts that he sends 10,000,000 emails a month to journalists, and inboxes are just buckling under the strain. And the thing is that these guys, they're aware of what they are that what they are doing is not wanted because they're frequently spinning up new nonexistent PR agencies with URLs that just go nowhere. Or well, what they do is they normally redirect to a real marketing agency, and they just keep spamming thousands and thousands of emails at journalists. And it's you know, they journal journalists are trying to, you know, trying to get the IT department to step in and sort of give them back their email inbox, basically. Shelby, you have thoughts on the scale of the scale of the problem? Yeah. I we would see it more from, like, a sign up perspective, I would say. Obviously, we're removing and warning and chatting with people, you know, probably 70 to a 100 times a day just from users that are allowed in the network. So like I said, assuming assuming best intent, maybe they didn't know that they weren't supposed to use AI. Maybe a client passed them along content that was AI, and we can kind of chat with them or or walk them through that. But even just on the front end scale of what we keep out of quoted so they never get in, I would say that's several 100 accounts per day. And that number can be kind of shocking, but we have to remember that that that could be one person who just sat at their computer. They have all of these tools, these ways to generate identities, ways to generate personas, ways to just generate a bunch of different email domains, and then just kind of continue to try to push and to get in the front door that way. So for us, it's it's a lot of keeping them out of the front door, which is the main thing. And then some people, once they are allowed in, we definitely have different mechanisms and tracking and ways to report. You know, our journalists can report and flag these individuals to us as well to help hold everybody accountable. And, ideally, that's either course correcting through education or it's removing due to this isn't what's right for our network, and this is what this isn't what we want in here. And I wanna go back, Rob, to something that you and I were talking about yesterday where I had said, I guess, in the scheme of things, some you know, when we talk about the shape of fakery, there's obviously what we've described a lot of today is sort of commercial driven fakery. Right? That is to say the goal of this is some monetary benefit, but the context is a much larger one in which you've got Russian disinformation farms. You know, maybe maybe the op maybe what they're trying to do is influence, you know, thinking around an election or something like that. And I said something to you like, I guess it could be worse. Well, But you your own perspectives about why even commercially driven fakery was was very dangerous societally. I mean, it's dangerous on two levels. On one level, it's directly dangerous when people assume medical expertise that they don't have and dispense advice that, you know, written by chat GPT or whatever that isn't very good at it. Dispense advice to publications that have hundreds of millions of readers because that advice stays there. And if you're dispensing medical advice and you don't have medical qualifications, you can harm people or even kill people. And, you know, there's a lot of these guys that dispense medical advice. But I think that more generally, there's this really severe problem in the volume. Like, one of the fake experts that we sort of unmasked on Press Gazette is a gardener called Fiona Jenkins who has a there's no picture of her even in the garden, and just has this sort of AI generated headshot. And she's dispensing very viral friendly advice like when is the last time to mow your lawn before before winter, etcetera etcetera, which hits for search and therefore is you know, it's gonna deliver as a piece of evergreen content. And the thing is that she doesn't exist, but she was quoted over a two year span a 170 times in the British media. And there is no real gardening expert who was quoted that number of times. And that's the problem with these guys is that eventually, the amount of fake AI generated content out there starts to swamp the real content. And then if you are asking chat GPT two years down the line for some advice, it's just reading its own advice. And there's this circularity to it, which means that the the information available for topics just becomes drowned in AI sludge, and nobody can get good advice on. Yeah. And thank you. I wanna make sure that we in a couple of minutes before I throw this open to the q and a for everyone else, I mean, clearly, the number one question everyone has is like, alright, Rob. Alright, Shelby. You've pretty well articulated kind of the shape of the problem, what types of fakes there are and fakery there is. You've you've talked about the scale of the problem. You've talked about the implications of the problem. What do we do to fix it? So maybe we should move on to before I throw it open to everybody on the q and a, you know, what do we what do you what you know, start with you, Rob. As a journalist, what advice can you give to other journalists to you know, who may be dealing with this problem every day? I think that it's it's a prob I mean, a lot of the problem stems from newsrooms in which people are told file 12 stories a day or you're fired, and where young reporters are often quite disempowered. So they're just given a list of stuff to do, which has been signed off by people much more senior than them. So there's no incentive for them to check whether something is real or whether something is not. I mean, the encouraging thing that we've seen is that we know a lot of news groups are actually taking this extremely seriously, including some that are global and have put measures in place where every expert that is quoted has to be you have to authenticate via things like a LinkedIn profile and a page on the company site. And for these guys who, as Shelby explained very, very well, they're spinning up these campaigns all the time. And they don't take a lot of the time, they don't, you know they the guy's not on the company website, and they may not even be aware of what they're doing. And there's no LinkedIn page, etcetera, etcetera. So and, you know, I know that several other news groups, this has been taken seriously at a very, very high level, and there are sort of reporters are now being encouraged to, for a start, not accept anonymous experts on anything and also to actually check whether people exist before they're printing expert led pieces. Yeah. So, Shelby, you wanna talk about some of the work that that you and the team do and the technology does at a at a Quoted? And I'm sure Quoted's not alone in in trying to tackle the problem. Yeah. So not positioning quoted is the only solution for this, but can you talk a little bit about what what you're doing? Yeah. It's it'd be so nice if there was just a way to go in and, like, hit a button. Right? And be like, is this person real? Yes or no? And, obviously, it's not that simple. So I think there's it tends to be like a multiple step approach. Like if we find something that's off somewhere, then we continue to dig further. And I think that's, I mean, one, it kind of comes down to your gut every once in a while. Like if something feels off, it's likely off. Let's investigate further. But there's a lot of things that come in. Like once you start looking at really fake looking websites and start seeing those websites, they start to become apparent. So then it's really looking at the expert themselves. And what type of social presence do they have, if any? Sometimes that social presence isn't always like, oh, they have a LinkedIn. They're good. Because anybody can create a LinkedIn. So now you kind of have to go to that extra level to say, like, is their LinkedIn verified? Have they put their government ID in LinkedIn? Does it have a clear verification, something like that? Was their LinkedIn just created last week? Because sometimes even they'll have several 100 followers, but those are also just all of the scammers following each other. So some of those those, like, initial immediate things where you're like, oh, yeah. They must be a real person. Those aren't always good indicators. So it's like once they start to stack up, that's where we will, like, definitely have to look at those things further. From I I I saw a question that was asking about, like, broadcast as well, but I think it's just even for real sources. It's the availability aspect and how available can that person be. So even if LinkedIn is there, I've checked it, it's not AI, but something still feels off about it, I advise people to ask them if they're available for a call. And I know there's so many journalists on here who are saying immediately, I don't have time for a call. Like, we do not have time for a call, our team does this all the time. We'll ask them for a call. Because nine times out of 10, if they are not real, they're going to say, well, we're not available for a call. How else can we help you to verify our identity? Or what. else can we provide you? We just we don't have time for a call. So that unwillingness to be either on the phone or in person or even willing to get on camera with you, that's a really good indicator. And most of the time, we are not at even having calls. It's just whether or not they're willing to get on there. So I'm not saying have a call with everyone, right. but. see if. available. Right. That that. changes the call even. if you don't have the time for it. That's an interesting take. If. somebody's like, oh, sure. Like, I could be ready in thirty minutes video or chat. It's like, great. I'm not this person seems legit. Like, I'm not saying you need to have a call with everybody. But if somebody can make themselves readily available to you, I think that is much more of an indicator that they are willing to be a real person for you as well. So Rob, yeah. I mean, were were you gonna say something? I was gonna say that chimes exactly with my experience as well. The it's they panic as soon as you go, can we just call? It's like they immediately are coming back with some outlandish excuse for why they can't get on a call. You know, I'm away on holiday or whatever. It it's it's there's always they instantly panic their backs up, and it it's different from what a normal person said. Because, you know, you might a normal person might say, you know, I'm on a flight right now, but I could speak to you in three hours or, you know, I can probably call tomorrow or something like that. Whereas when it's a fake, they're immediately like, hey. I I can't ever call. Or making. up really strange excuses for it. Mhmm. Shelby, could you just take a minute to talk about you talked about what you as on the and the human team inside of Quota do to validate. Do you wanna just talk a little bit about GPT tracking, vetting, flagging, some of the technology that exists? And I'm I'm again, I'm sure it exists elsewhere in other platforms, but you're obviously gonna be talking from the QuotaD ex expertise. What sort of tech what what can technology do to help? Right. Yeah. Like I said, I don't think you know, like, if I do a AI check on the content and that is you know, that's coming back as AI. I'm not immediately saying like, hey. Completely disregard this person. They're not real. I think it's the accumulation of multiple things sometimes. Because like I said, we want to assume best intent. Right? We're still trying to be in this world where we're like, hey, maybe this person just didn't know how they were submitting it. I think location is a big indicator. So we definitely want you to take that into account, where you see where the pitch or the message is coming from versus where that source is from. There tends to be, like I said, a lot of offshore resources submitting on behalf of your sources. So location for me is huge. That's something that we we definitely look into. Even if it's just your executive assistant submitting on your behalf, it still feels weird, and it still feels like it's probably not content that's genuine and coming from you. There's, of course, we're partnered right now in Quoted if you use Quoted. But there's plenty of AI detection tools or checking tools where it's looking at your natural language processing. The best one that we have found and that we have adopted as a company is called Pangram. The reason we adopted that one is because it, right now, compared to anything else in the world, has the least amount of false positives. So it will give you a breakdown of those certain sections that appear to be AI. It is the best in the market right now. Who's to say, as AI adapts and grows and changes, that we may also not have to change into another platform or start utilizing other tools in the future that continue to combat this? But as of right now, Pangram is the best one that we have found to be able to align with, and it is very, very accurate from what we've been able to see. And it highlights those areas of the content itself that are indicating being fake. On the other side, AI is not all bad. Right? So I'll often I'll pull somebody's persona, and I'll plug it into ChatGPT. And I'll have ChatGPT do the work for me and say, can you pull me details about this person? So if they're saying they're a lawyer based in Louisiana, something like that, can you find anything validating that? And it will tell me if they have a registered license with the state, if they have any of that information. So use AI to combat AI and ask them if it's real because it will give you the pros and cons. Like, hey, This person has a website, but I can't find them registered anywhere. They actually only came into business last year. Something seems off here. AI is the first one to tell you. And I have found that that can be really, really helpful when you're kind of plugging and playing some of this information and and messing around with it. But then on our side, we do want to continue to kind of police the network. And so if Rob has somebody spamming his inbox, I don't want that person coming and spamming me. So my hope is that Rob reports that person and kind of flags it. So while we police this, I quote it as much as possible. And we are educating these people, trying to course correct, and removing the really bad actors from the network. The more we can kind of flag those to one another, the better it kind of helps hold the entire network accountable and keep those people out of here to keep the best sources and best individuals in here pitching you so that it's helping stories long term. Yeah. I mean, I'll weigh in with my own sort of editorial position here, which is I think the answer to this is networks. And I think networks get a bad reputation because we think network you think social network, you think x, you think bots and trolls. But, actually, I use network when I think about things like Uber. We get every day, we get into strangers' cars, and we trust that the network and the internal vetting and all that other, the flagging and bad actor flagging and verification from the company itself kinda keeps us safe. Likewise, kind of always boggles my mind that I can or I eat food that strangers sometimes deliver to my house, and I just sort of tuck into it without really thinking about that. And, that's the same thing. We're trusting the network to kinda keep us safe. So, you know, obviously, I quoted you know, our our contention is that a well regulated network with the right rules in place, you know, will keep will help ameliorate this problem, probably not solve it completely, but ameliorate this problem. I wanna make sure we ask ask some very specific questions. So, for example, Rachel asked you, Rob, were the articles that were analyzed presented as true editorial stories? So in. that Absolute yeah. Absolutely. They were I mean, there's there's a couple of them where they looked a little murkier, and we wondered whether there was a separate story there that publishers were running questionable content in order to drive affiliate revenue. But in most cases, these are running a straight editorial. I mean, in a lot of cases, they're running on news websites where they have an extremely high output of kind of lifestyle, so viral friendly content. And, I mean, there's one news group where one guy output a 100 stories in one day. So, you know, they are clearly using AI to to create these stories, and they're not very fussy about where the stuff's coming from. So it is editorial. It's not the highest quality editorial, but it's editorial. Got it. So other questions that we had. Does Quotid have an official policy stating that expert content should be never be created or helped by AI? Shelby, do you wanna take that one? Yes. We do have a an internal, like, terms and conditions and an an AI policy. We've we've written it sort of loosely, so it's not just like one strike, you're out. Because like I said, we do wanna assume best intent where, you know, I have somebody in here who's a business owner, and they just didn't realize that they're not supposed to take their thoughts and have ChatGPT rewrite them for clarity because that reads as AI. So we're helping to educate those users. But having that policy in place does allow us to remove people from the network. It's a bit of a sensitive subject for me right now just because those people that we do remove, obviously, they're the the angry ones who get quite feisty and argue with us and, you know, kind of berate our support staff all day long. So obviously, that's a good indicator that they're not meant for the network if they need quoted that badly. But they're the ones that then turn around and give us, you know, like, one star reviews on Trustpilot and things like that. So we're actively removing them from the network. If you have any questions about whether or not we're removing people, you can go look at our reviews because people are very, very angry at us all the time. And the and the way we do that is when they're flagged internally or if we have internal safeguards in place that are flagging these individuals, we chat with them. We will push them to have conversations with us. So we're having hundreds of on, like, on phone interviews several times a week throughout or throughout the week with our different support staff members. And then we're also having just chats back and forth to kind of get a better understanding of who we're dealing with, and then we make the determining factors from there. That they're allowed to stay within the network kind of on a preliminary basis or on a conditional basis. Amy says if you if they're angry, you're doing your job, which almost all of our team have a, like, a plaque that pretty much says the same thing above their desk. We've got about five minutes, and I wanna touch on a question here from Jennifer. And I'll take the first stab at it, but Rob and Shelby, you you touched on this a little bit before. I don't know whether this is Jennifer maybe didn't catch the beginning, but it's it's sort of worth underscoring because it is a good question. It boggled our minds, you know, at the start of the sort of the GPT craze. So Jennifer asks, if a source is fake, what's the motivation? Sources aren't usually paid, and if it's not a real person, then they're not gaining fame from being in the article. So why do it? And I'll take the first answer to this, and I'll throw it over to you, Shelby and Rob. But, I mean, the answer that we understand is is search. Initially, search engine optimization is the way you'd show up in something like a Google, and then more laterally search as in GEO generative engine optimization, the way you might show up in a chat GPT. All of this, whether it's Google's original search engine or something like a Gemini or a or a chat GPT, is pulling from the web and optimizing for certain content. And, you know, one of the ways that it optimizes is to say, well, did a reputable venue quote this person from this company? And if they do use that, then they're driving eyeballs to at the end of the day, possibly a website link. And so if you spun up a fake website like, you know, drycleaningtips.com, the value to you is that you're growing the eyeballs to drycleaningtiptips.com. You can, you know, eventually monetize those eyeballs by advertising against them, and you can eventually maybe even sell the site for a lot of money. So, actually, there's a vast financial incentive to doing this, which is why it's such a big problem to to stomp out. And then, of course, you've got the political kind of trying to influence, you know, hearts and minds. I read an article just yesterday about the number of TikTok posts in the Ukraine, something like 80,000,000 that were trying to paint the Ukraine war as an American proxy war. Right? So you have that incentive going on as well. But it it certainly on our minds when we first encountered it to be like, well, what is the monetization mechanism? That's it. So, Shelby, do you wanna sort of or Rob, do you wanna elaborate on that? Did I nail that, or would you clarify it? Absolutely. I mean, it's in these cases, you know, it's it's it's just SEO. It's the links that they're after. I mean, I I come from the world of print, so I'm always baffled as to why publications even put in links because but it it's it's those links. And just to touch on one other point, the the idea that experts wouldn't be paid. I mean, if you've got somebody like a psychiatrist working for your organization, they would normally be paid. What these guys are doing is they're creating fake experts who are obviously free to get generate the kind of reach that you would normally have with a PR campaign that whether you would have enlisted an expensive celebrity or expert in order to provide that authority to your brand. These guys are faking it and doing it for free. So we are coming up right on time, and I do wanna you know, they say sort of leave them wanting more. So I think this is a good opportunity for us before this gets cut off to thank Shelby. Thank you, Shelby. Thank you, Rob. This has been super helpful. You know, Rob, I'm assuming you're gonna continue to keep reporting about this, Well, there's more to come. right? And and and, Shelby, you're available inside the quoted platform for users to to reach out to you about this. So we can continue the conversation, you know, in the in the following, you know, months and over this year. We're not going anywhere. I don't think this isn't going anywhere as an issue. We'll probably end up doing another one of these sessions towards the end of the year. So I wanna say thank you to you. I wanna say thank you to everybody that joined. And, you know, I'd say keep it real, and that would probably be more more more apt than ever. Thanks, Yeah. Brilliant. everybody. Thank you. Great. Thank you. Thank you.