Fake AI ‘experts’ – how is Reach hitting back?

[Music] Hello and welcome to Beyond the Noise, the podcast series from PL Week. I'm John Harrington, the editor of PL at UK. Our special guest this week is David Higgerson, Chief Content Officer at Reach, publisher of the Muir Daily Express, okay magazine and many more titles. Hi David, thanks so much for joining us. How are you today? I'm good, thank you very much, thank you very much for having me. Pleasure, nice great to have you here. David is here to discuss a growing problem for publishers, namely fake AI experts, as the press Gazette has been reporting in recent months, several media titles have inadvertently quoted apparent experts in articles where they were laid to exposed as fake. Reach is one publisher that has been acting. We're going to hear about that today, but first I thought we could have a hear a bit more about David in recent developments. Reach, so first question and a simple one, what is your role in tale exactly? So I'm fortunate enough to have the job of leading all 1800 of our journalists in the UK. I'm responsible for the journalists that we create and also trying to plot the future to ensure that journalism is sustainable going into the future. Brilliant, well it must be a fascinating job at the moment, obviously, with so much change going on. I mean what would you say the kind of the biggest ways in which Reach has changed its approach to digital content in the past, say sort of 12 to 18 months. I was looking at the financial results for the four year recently and there was really a huge growth in specialist video roles. I wonder if maybe you could talk about that a little bit. Yeah, absolutely. So for a long time we've focused on maintaining our scale. We still have that scale 35 million people in the UK, but last year we introduced the new strategy and around new CEO peers north to focus on really ramping up our video, both in terms of the big set piece productions such as the the daily espresso that's the room I'm in today and also making sure that we're telling as many of our stories as possible through video and pushing it into the places where people are consuming that content. So our focus is shifting from where we were previously, which was trying to get as many people to come onto our own operated platform. That remains incredibly important to us, but at the same time putting a much greater focus on finding people where they are and serving up the journeys and the we create in the formats and ways that that people are telling us they want to consume it in. Yeah, that's really interesting. I mean it's certainly something we've known as we've noticed as a publisher at Haymarket, the increasing importance of video. Do you find that a lot more of the links back to the website and so on would come from video clips and it's very important as a promotional tool as much as anything else? I think clicks back to the website after people watch video are great. For us it's more important to be serving up the journeys that we're creating where people are growing awareness of the brand. So like you say, as a promotional tool, in some cases, that's introducing people to our brands and to specific individuals within the organisation for the first time, for example, on TikTok. For others, it's about we have more than 100 million followers across all our social platforms, Facebook's the dominant one there. We've been very good at being a Facebook publisher for a long time now. So we've got a lot of people who are familiar with our brands but are choosing to consume content in different ways. And that's why we created the everyday journalists and video teams who are based on the mirror, they express the star and around the UK and Ireland. So every morning with when we come into the office, we're thinking about what we're leading the homepage with, what we're going to leave the newspaper with, but having that conversation in the newsroom about what is it that's getting people talking and watching on social media and design news agenda reflect that. And if it doesn't, what are we going to do about it? Brilliant. Well, I want to talk about AI a little bit and how that's been used by editorial teams. I mean, it's something that, you know, every publisher as much as every PR agency and PR professional has been thinking a lot about how has it been used by editorial teams across each? So our principle from with AI for the start has been how do we use it to make the job of journalists better? And certainly as we've been on a journey into digital publishing over the last 10 years or so, we've put more and more expectation on what our journalists are expected to do. And without sounding like the old man that I probably am in danger of becoming is 20 years ago when I was a reporter, 25 years ago when I was a reporter, you wrote the story and that was the end of that. With these days, when a reporter writes a story, they often have to source the image for it. They have to write the picture caption for it. They have to remember to put the links in the articles. They have to think about the SEO headline. They have to think about the social headline. Obviously, they're supported by content editors and others in that process. But we were really keen to find ways that enabled us to tell more original journalism. So removing duplication of effort and also finding AI tools which enabled us to focus people on the parts of the journalistic process where journalists add more value, which is the storytelling and the story projection. Our main tool is a tool called Guten. That enables us to take a piece of content which perhaps was being written in 15 different locations because it was an interest to 15 different audiences. And we now write it once, but we can tell that in 15 different ways, which gives much more tailored journalism to each individual title and indeed, freezer resource to focus on more original storytelling that's not been told elsewhere. We also use tools like Trint which speeds up the process of transcribing press conferences, enables us to create more content on the back of that. And it's also widely used in the audio video process as well, again, freeing up people to really add value to the content generation, the ideation of what we want to do and then spending more time talking to audiences at the end of it. What sort of reactions you get from the journalist to these sorts of products? I mean, I'm conscious that part of the sort of journalist mind might be a little bit averse to these sorts of things and it's a conversation we have at Haymarket, the publisher of PR week, quite a lot, but at the same time, clearly they're very useful in some ways and in many ways. So yeah, I'm interested to know a little bit about the kind of reaction and how useful they are at the moment. I would say sometimes journalism gets a bad reputation for not wanting to change or embrace new things and that may have been the case years and years ago. I don't think it's the case now. In the early days, I think that first wave went sort of chat GPT arrived on the scene and the flood of early tools. I mean, our challenge was more making sure that our journalists were using the right new tools, so ones that they could use safely, because there were some tools out there which you promise to do all sorts of things, but also involved lots of data and information being shared that perhaps the journalist also, or we as a company, didn't want to share. So the enthusiasm for AI adoption is definitely there within our new dreams. Is there a nervousness within the industry and within our new dreams about how it could be used? Absolutely. And we've been very clear about how we want to use it. It's not there to replace the story creation and the storytelling process. It's there to assist it. There are lots of ways that it has been used and new use cases are being proposed by journalists all the time. At the same time, it's possible to also be very, very concerned about the effect that AI is having on wider journalism or wider content creation. And there is no doubt about the fact that we do now find ourselves in a daily battle with an awful lot of AI slot that's been produced by non-journalistic organisations that is competing and sometimes beating us in Google. And that's something that we're pressurising all the major platforms to take seriously. Interesting. And you mentioned Google and obviously a lot of the publishing industry has had quite a lot of a dislike for Google's AI and the AI summaries that it's been argued take a lot of traffic away from websites. How much of a problem do you see that across reach? I mean, there was some data there that looked a little bit concerning in the results. Was it 46% you and your reduction in traffic from Google? What impact do you think that's having at the moment? So there are two sides to that. So a lot of the 46% that Pears talked about on Tuesday was driven by Google Discover. Now, Pears is absolutely right when he says that when we have the opportunity to get a lot of traffic, we should absolutely take it. And that's what we did with Google Discover. Google Discover changed its focus in the midpoint of last year. It onboarded more publishers. Unfortunately, it let a lot of AI slop in as well. It's seeking to address that. So the bulk of our decline in Google has come from not having much space on Google Discover. That's not to say it's not still a really important referer for us. It is. And we do have a constructive conversation with Google. AI Overviews has been the challenge for us as a publisher which has focused heavily on driving audiences from social over the years. We have not been as dependent on organic search results. Therefore, the impact of AI Overviews hasn't been as pronounced for us as it has been for other publishers. That said, that's not to say it's not a huge challenge for us because what we do see is the relationship or the trade that publishers have always had with Google, which is in a nutshell, has been they serve up our content. People click through to our websites to see it. We monetize it there to fund that content, which ultimately helps Google be better. Has been broken by AI Overviews because it is reducing the likelihood of people clicking through. We're very much a breaking news organisation. AI Overviews tend not to trouble that space too much. But if you're a editor of said Chronicle Live in Newcastle, it is very frustrating when you are paying a what's on writer to produce brilliant guides to where to eat in Newcastle and then Google then chunks that down. a few paragraphs in an AI overview and we don't see the monetary return from that. And that's the conversation that we're having with Google. It's the conversation we're having with government because ultimately Google needs to be a better partner. But it is fair to say that Google is actively engaging in conversations with us. The same can't be said of all the AI platforms, all of whom are essentially doing the same thing. Google is a more difficult case because up until now we haven't been able to separate our content being scraped from AI overviews, from our content being scraped to be certain in Google search results, which continues to be a major source of traffic for us. So it is a difficult position for us to be in, but conversations continue with Google and Google are talking to us all the time. A bigger concern for us with Google on the search front is that we see on a regular basis our content and the content of other publishers being squeezed out in favor of content from less reliable sources. And that's something that Google are taking seriously, but we've yet to see that problem solved. That worries us both in terms of our journalism being seen and also the chance of mis and disinformation being seen first, which is a real issue for society as well. Yeah, absolutely. Well, thank you for all of that. I'm going to move on now to the kind of the main thrust of this episode. When would you say reach first became aware of the problem of fake experts and how bad is it now would you say? So, PreskZet brought it to our attention when they talked to us another other publishers a while ago. We responded quickly because ultimately what AI experts or fake experts rather is doing is very quickly breaking the bond of trust between us and our readers. And in some cases breaking the bond of trust between journalists and their contacts as well. In terms of how bad the problem is now we're confident that we have got an increasing number of measures in place which ensure that our journalists can verify whether somebody is real or not, which to be frank five years ago if you'd said to me that our journalists were going to have to spend time with press releases coming in and job number one was verifying whether the person was real or not. You'd have laughed at it. It's still a problem. It's still out there. And I think it speaks to the problem with that I guess AI has introduced into our world which is that the host always been people out there who've wanted to gain the system for their own ends. AI tools make it far easier to gain the system and it's our job to try and be on the front foot as best as we can. What would you say some of the more egregious examples of the fake expert phenomenon really? The point where it became really egregious for me was if it was as simple as do a search and see if the person exists and they don't appear to exist you would say that we can solve for that. What we've seen over a period of time and as press exists and it's reporting and as we sort of responded internally as well is it's the length that some of these actors go to mislead. So it's not enough these days to check and see if somebody's got a linked info because that can be faked. It's not enough to see if they've got a Facebook profile and the length and depth that people will go to to try and create the impression that somebody is real is what I find particularly egregious. The examples from press Gazette are well documented. They tend to be in the space of lifestyle in terms and that's the stuff that there's an essential appetite for from readers which what are the life hacks that can make my life better. The challenge for us is every time we put a new safeguard in place and those safeguards are really about helping journalists do the verification quickly rather than taking away the need for the journalists to do the verification is we see the need for us to put another step in or respond to something else. It feels like a war of attrition at times. Yeah I could imagine and where do you think they're coming from? I mean really the press Gazette article does sound like some sort of shady organisations frankly that are ascending a lot of these. I guess I spend with my colleagues less time in trying to understand the why or the specifics of the who and more time on how do we build tools and processes which mean that we're not losing a lot of time to having to verify whether a person exists or vouch for their qualifications. Obviously it's really important that every journalist does both of those things but how can they do this as quickly and as efficiently as possible. I think the thing that probably worried us the most about this was if it was a case of it was a you know a PR agency, a PR firm, an individual that we'd never heard of before you could say right that self-contained to this one company or this group of companies. I think the thing that probably spooked us more was when we started seeing some material that was being flagged as being fake experts from organisations that we have very good relationships with and when we spoke to them it was a case that they had also themselves being jubed when seeking material to support their press release output. I think that showed us at that moment just what the size of the challenge is here because I guess we're facing the age-old problem that with every new technological development comes the opportunity for somebody to make a quick book and it's up to us to make sure that we are still doing right by our readers and coming up with the right safeguards as a result. That's fascinating and quite disturbing. Do you have any more details on you know those sorts of scenarios where the PRs themselves are saying they're being jubes? Well as I understand it is and again the focus of our attention has been on you know how do we make sure that we've got the right safeguards and checks and support in place for our journalists. Whenever we have found an example or take the press Gazette list for example gone through the articles in question and not all the articles that necessarily have been flagged to us initially have been found to be fake experts that I can talk about some of those examples in minutes well if that's helpful is we will go to the PR firm or sometimes the company directly themselves and say what's going on here and it is people who have you know behaved in good faith people who got longstanding relationships with who they've acted in good faith with a partner or third party to get information or a quote to support the press activity that they're working on and they've ultimately been they've ultimately been misled. There's a view that I'm sure some journalists would have which is well this is this is why you should source everything from scratch yourself not rely on PR and in the case of PR agencies not rely on third parties to help you assess and build what you're trying to promote. But the reality is we have a system of a relationship between the PR industry and journalism which has worked very well for a long time but as happens in all relationships from time to time bad actors can enter and and cause challenges and certainly where we dealt with PR agencies who themselves have found themselves to be juke they've been very quick to respond and of course are very very keen to maintain a good working relationship with us. It'd be good to talk now if you can through some of the steps you have been taking to solve this problem. Yeah so the thing that seems to have caught most people's attention was the centralized database that we've created at reach which was well it's probably easy to say what it's not because it seems to spend a lot of time talking about that on LinkedIn. It is not us trying to determine what a good PR company looks like or to have a kind of if your name's on the list you can come through type approach. It's about us having a list of PR companies and that can be one man band through to the biggest companies the individual journalists in our organisation are happy to vouch for happy to say they've got a good working relationship with them they trust them that provides all 1800 of our journalists in the UK a quick way to check if they get a press release from somebody or contact from a company that they don't know they can very quickly say do a quick check to see if that's endorsed and supported by a colleague and that helps them move through to the next step of the verification process. We've also used using AI so to a certain extent using AI to fight AI. Using Google Gemini we've created a gem called the detective where you can put a press release or through and it will tell you the likelihood of the person in that press release being authentic or not by doing things like checking their LinkedIn profile, checking their LinkedIn profile in a way that makes sure that it's not just a very sort of superficial couple of posts and a AI generated image. Looking at where else they've been quoted and sourced and then giving that a score out of 10. What it's not about is saying to the journalist who's producing a story therefore you can be absolutely certain this person is who they say they are the heart of journalism sits the need for the reporter the writer to know who they're quoting, know who they're speaking to and to build an article having the new checks. The challenge with what we encountered was that there was a danger of overwhelming our journalists with the amount of time they'd have to spend on verification of individuals so we focus on building the tools which make that process as quick and as efficient as possible. So how many is this the right way to put it but could you give an idea of how many sort of organisations or individuals have been quite a quite sort of approved as part of this? Yeah, be good to get an idea of the kind of scale of this this operation that you're doing. So we've, our AI team have worked with external lists and have pulled together a core list which is in the many hundreds of firms. Islandvil yangen garr karateg I am Aodgenwaidd yn ddergymr ar Ew動ad yn pen unor collen hangwedig rhyw am barhun petog h那我們 peuli feiain Clewanicau beryrdiod dit y rhoeddan ph heritagead fwycry pgeud fel mae fel wedi ddi nhw gweithio bod fydd hyn problem Kaus am gyda'r gyda'r dyma fedreool am er eng a Faiftlyg, yullر cy— carbono am Mentor Uain Emanuel Cefrán Tall. A sef y Kith, cynnod, yn gallwn ni hanes ffermaid o'r lazy o gyfryst o ll Rainbowll. Yearn lleol hynny ceff hairf spoken byd eru Gymod yn castol o'r ysanger. Ond yn y ceff hairf singol ei gwyfl nad daban wedi lidau Oesmorid yr oeddion gy a compress y gallwn na i feb totr fidr HERD. fod. N â'r »6 VoFolwyr mith cast oedg eu reín einig'n glawnarr. Mae'n hiw hosteg s moisturizing Ew alle nhw yn ad900 eït ein mynd i gerdillio ffordrinaidd ac o rybau. Ysglad o gwneud owerth i ycular o gw budgetorol gyffreith yr elio huge na다 aparech ond mecribed. Thysdydd mam'n gwyli fixed, erbyr加pt sy'n y nien, cyn же feithio iawn, cyf dabeiolioli gwrs Ștydiolio, yna Ysglad. Yd yna etoangled, cerat ' أ nwylach yn cael ei defnoeg, yn yw'r yw'r plafor ac yw'r ysglad ydy'r plafor yn y ffyrraid yna o ddod yn ffyrraid. Ysglad yna, ac yna o'r llyfr yw'r yw'r ymwyr i'n gael ei ffordrinaidd. Mae'n gael, yna yna eto yn ffordrinaidd o ddynod. Mae'n gael ei ffordrinaidd o ddynod yna eto, yna eto yn yng Nghymru, yna eto yn yng Nghymru, yna eto yn yng Nghymru, yna eto yn gael ei defnoeg, yna eto yn gael ei yna eto yn gael ei ffordrinaidd o ddynod. Mae'n gael ei ffordrinaidd yng Nghymru, yna eto yn yng Nghymru, yna eto yn yng Nghymru, yna eto yn gael ei ffordrinaidd. Mae'n gael ei ffordrinaidd yng Nghymru, yna eto yn gael ei ffordrinaidd. Mae'n gael ei ffordrinaidd. Of what it means not to be on the list that we're operating. The solution is to have the relationship with a journalist in our organization and happy to vouch for them. That's all that we're trying to do here. But you know, we've got 1800 journalists, 1,300 of them are focused on story creation across video, across text, across still imagery. They're always looking for great stories to tell and that's the relationship, that's the relationship to foster. Where we have also generators at some interest is that if we do find repeat offenders who are supplying stuff, which is misleading or designed to be disingenuous or designed basically to exploit the relationship between us and our readers, we may well take action to say to the entire editorial function that be hyper aware with stuff that's coming from this firm or this email and as has been previously reported, we may well take the decision to block their material coming into the business. We're not there yet in some ways I see a read across between the days where people running the backlinks at agencies and the dark arts surrounding that would go to any lengths to get backlinks on our websites. So we're following a pattern that we've worked on before, which is about protecting our journalists and ultimately protecting our readers as well. Right, what kind of discussions are you having with the sort of PR industry outside of the kind of day-to-day discussions you'll always have with individual contacts from the journalists? Are you in discussion with, for example, trade bodies or any other authority groups within comms about dealing with this problem? Well, more than happy to have those conversations. We haven't set out to have those conversations at the moment because as I said, not in the market to become a kind of sort of de facto regulator of the industry or even a barometer of what good or not good PR looks like because that's not what we're trying to achieve. I've spoken with a lot of PR professionals in the past few weeks from big firms, small firms, public sector and different parts of the public sector. And I get that when you see the headlines from us which says, "Reaches attempting to audit the PR industry or reaches operating a white list or a black list," that are absolutely get why some people have worried the way that they have done. But hopefully whenever we do talk to people in detail about what we're trying to do, people get what we're trying to do because ultimately what we're trying to do is protect the relationship between us and our reader. And that's a relationship that when we work, we as journalists do with our PR contacts works well. That enhances that relationship between the reader and ourselves. Right. And are you taking any other steps to look at this problem or do you imagine you might take other action? Well, we're constantly looking at what more we can do. So what more can we do to make the process even simpler for our journalists? You know, the utopian idea is that you receive something that looks like it's coming from a PR firm and you can trust the person exists, you can trust their credentials etc etc. But we're not in that place. And there are some people who have said to me as well that we're using a sledgehammer to crack a nut into a certain extent. And we're trying to do that because we produce north of 3000 pieces of journalism a day across reach in the UK Ireland and in the USA. But it doesn't take many of those to be focused on as having a fake expert in them. And for people to become concerned about our journalism. And that's why we have taken this as seriously as we have done. And we don't just take it as read when we're told that somebody is a fake expert that they are a fake expert. We do investigate that ourselves. And indeed there have been some of the fake experts that have been put to us. We took the conclusion that they weren't a fake expert indeed one was one of our own members of staff. So we're pretty confident they weren't a fake expert. Do you have any final word for the PR industry? Because I'm sure you know I'm spoken to a few people about this in the sector. They're very keen to work with reaching other publishers to deal with this problem that seems to be as much for a problem for legitimate cons professionals is for publishers really. Any final messages or things you'd like our listeners to think about? Yeah, I think in the main fact the rules of the relationship haven't changed. You've got stories you want telling we've got stories we want to tell it's where the relationship is and how can you help us make sure that we're serving our readers with reliable accurate information that our readers care about. I think there's been the transformation in digital journalism over the last 10 years or so has transformed the way we understand audiences and everything that we do is based on wanting to get our stories in front of as many people as possible and in particularly in front of the people where it can have the most impact and. The fake experts thing is not good. There's no two ways about that. But in the same ways we would hope people recognize that it's a very, very small number of stories that it's involved reach and indeed other publishers. We also recognize that the vast majority of the PR industry is trying to do what it's always done which is work with us on telling great stories and that's why everything that we have done so far is designed to ensure in our journalists can spend their time focusing on telling great stories and where appropriate work with PR professionals as has been the case for many many years. So I definitely don't want your listeners and readers to think that reaches somehow empty working with the PR industry we're all about telling great stories and we want to continue doing that. Brilliant. Well, I think we're just about out of time but thank you so much David. I feel that's been like absolutely brilliant insights and I'm sure our listeners would have learnt a lot as well. So thank you and thank you to everyone for listening until next time. Goodbye. [Music]