Mathias Döpfner, CEO of Axel Springer, the largest periodical publisher in Europe—and an increasingly major presence in American media—is trying to adapt his company to the new reality of artificial intelligence. New tools such as OpenAI’s artificial intelligence chatbot ChatGPT have created opportunities for media companies to cut costs by enlisting technology to do at least some of the writing, research, and copyediting. But at the same time, A.I. threatens to steal the jobs of reporters and create error-riddled articles.
Axel Springer recently culled 200 jobs at Bild—20% of its newsroom—at least partly because those jobs could now be performed by A.I. or automation. The layoffs were a step toward Döpfner’s “digital only” vision, his strategy for ushering in the A.I. era of journalism at Axel Springer and maintaining profitability.
Döpfner, whose company includes many popular multinational news outlets, including Politico, Business Insider, and Bild, will have a big influence over how A.I. shapes media.
On June 21, Fortune editor-in-chief Alyson Shontell interviewed Döpfner at the Cannes Lions Festival about the impact of A.I. on the news industry, related developments at Axel Springer, and predictions for the future of journalism.
This conversation has been edited and condensed for clarity.
Fortune: You have not been shy about saying how A.I. is transforming your companies already. Just yesterday, there were hundreds of layoffs announced at Bild. In the memo, you said these roles are jobs that really could be done by A.I.—and with more to come. Already over 17,000 jobs across media are gone this year, as one of the steepest years of cuts in the history of journalism, that really hasn’t even been impacted by A.I. yet. What do you think the future of newsrooms looks like? Will it just get smaller from here? You talk a little bit about which roles might be protected, but which do you see really withstanding A.I.?
Mathias Döpfner: We can be very concrete, because what we have announced yesterday is exactly one example that illustrates what I just have tried to explain. And that is that we don’t want to be forced by developments that are so obvious everybody does it—we want to be ahead of the curve. We want to be at the forefront of these innovations, and we have said it is, for us, very obvious that certain structures and certain workflows are just not up to date any longer.
However, those outdated production structures and workflows should be changed proactively. And that means concretely, in the case of Bild, what we announced yesterday, we have 18 different regional offices, big physical offices, with PAs, with office heads, with production managers, with the central layout, the central production editors, and so on. And we just think that is the wrong allocation of capital and of people. We don’t need offices in the physical sense—we have remote work. What we need is reporters on the street. We don’t need decentral layouters that can be done with the help of A.I. very efficiently and more centrally, but what we need is more investigative reporters, more experts who are reporting on certain topics, more great writers.
Basically taking out some jobs here, and as I said yesterday, doubling the number of investigative reporters at the same time, I think, is going to improve the quality of regional reporting. Now, some people say: “This A.I. is just an alibi. They want to cut costs.” No, I can tell you that’s not the idea. And we are doing that out of a position of economic strength and success. Fortunately, particularly our German media business is going better than expected in America. We also face these headwinds of digital advertising trends that are not so positive at the moment, but in Germany, the market is 20% down compared to the previous year, and we are ahead of the previous year. We are up, so we are doing it out of positions of strength. We do it mainly, or I would even say only, because we are convinced those are the structures of the future.
I think you’re referring to the ambition to have $100 million in cut costs by 2025 across all of Axel Springer. And you’re right, in the U.S. it’s a different circumstance than in Germany, although this memo did clearly say that editors’ jobs, photojournalists’ jobs, those jobs will be permanently changed moving forward. Another thing that will probably be permanently changed, definitely, is the business model of media, which has been incredibly difficult over the past few years, even with social media and now A.I. What does A.I. do to the business model of media? What happens to subscriptions? What happens to programmatic ad revenue, to direct revenue, to everything? Can publishers really survive?
First of all, how exactly will things change and transform? I don’t know, and I have no crystal ball. I have some assumptions. I think that advertising remains very, very important. I don’t see a reason why it wouldn’t. But how advertising is executed, that will change. It will be more efficient to the benefit of customers and to the benefits of media brands. Subscription will get even more important. I think also the whole definition of which information is really needed, and which information should be worked on by human beings needs to be clearly distinguished.
I think the whole idea that we aggregate information that is out there because wire services have provided it, or other media have reported on it, and that then an editor sits down and rewrites it, is a totally outdated model. Nobody will ever do it. You do it in a second with generative A.I. When large language models can simply do it better, why would I need a person for that? But the person who’s really writing the super interesting, fascinating, charismatic, contrary editorial, or the person who is coming up with a great report on Cannes Lions at the beach, that will be always needed.
You’ve said only the best original content creators will survive, which sounds exactly to that point. What role do the platforms play moving forward, do you think, for publishers? We’re seeing how one change from a Facebook algorithm in the past few weeks sent a lot of traffic across the internet plummeting. Obviously, with generative A.I., Google and these companies can start making a lot of content themselves. As you say, aggregation is no longer needed. That goes for a lot of evergreen content, which has a lot of journalism jobs right now. How will publishers attract people to their content if they can’t rely on platforms?
First of all, it’s interesting that two big platforms are main players here, too—it’s Google with Bard, and it’s Microsoft with OpenAI and ChatGPT. And they will keep very, very relevant positions. The likelihood that this time they have a very, almost egoistic interest to help publishers, creative companies, content creators, to keep financial incentives, to keep business models, is very high. Why? Because they need what we produce. And they know if they don’t do it, the likelihood that they will end up in trouble with politicians who then discover that as a threat to democracy and threat to the economy in general is so high that they would rather proactively reach out and say, “Okay, what’s the best model?” And we are saying a quantitative model like in the music industry, where first of all, they have to prove which IP they crawled—not we have to prove that our IP has been crawled.
That’s the main goal. And then if there is transparency of what they use, there will be certain methodologies of how you quantify, and according to the quantity of IP you use, you redistribute license fees like the music industry. If the music industry didn’t have that since the ’60s, there would be no single record label or the whole music industry would be dead. So I think that is a kind of analogy. But there are also other models conceivable. But I think the interest is very much also on the side of the platforms, and that’s why I’m more optimistic for the future of media business models. Actually, I think they can improve because you get lower costs and a better product with more interest in remuneration.
Do you think that publishers will need to create their own walled gardens to protect against the crawling, or their own chatbots? I mean, are we expecting people to start coming directly to websites if you can’t rely on the Facebook referral traffic or the Google referral traffic anymore?
Well, there may also be different solutions. I do not necessarily think that it needs to be a walled garden, but there should be healthy competition between different solutions; perhaps some top U.S. players, some more innovation and activity in Europe. There are some interesting startups and companies that were doing LLMs [large language models], but I think more to come. And then every publisher can decide. Some may want to go for a walled garden, others don’t. I think there’s not only one model that fits all.
So Axel as well as other publications have been working on copyright protections, and going out to the platforms to try and say, “Hey, pay us for what you’re crawling.” How are those efforts going? And one thing that you said that was a little bit jarring to me was that you felt like if copyright protection is not successful for publishers, there is no business model for media moving forward.
I would even go a step further and say that copyright as we know it is already gone. I don’t see it. I mean, in this world, the whole idea and the definition of a copyright just doesn’t apply. So we have to rethink the idea of copyright, and that is important. If the old idea that intellectual property and data and content that people have created—whether it’s a movie, or whether it’s music, or whether it’s literature, or whether it’s journalism—if that is something that is absolutely not protected, everybody can take it and use it, then there is no business model. Why would anybody invest in that? Nobody has incentive—you just steal it from other people and use it and monetize it. That’s nice, but that will deteriorate the quantity and quality of content so fast. I think as I said, it is not in the interest of the big players, the platforms, it is not in the interest of society and within the interest of politicians, and of course, it’s not in the interest of companies. So that’s why I think we need to figure that out.
And how receptive is an OpenAI or Google Bard to copyright protection?
They used to not be very receptive, or with certain nuances. You could say, one platform was less receptive, the other one was always a bit more proactive and understood the needs. But I think in this case, they are all pretty receptive because of their own egoistic interest, and also because they fear that if they don’t, the price they have to pay us could go like in Italy, where it’s just forbidden. That’s a theoretical option. I don’t know if that’s the right one.
I’m also not a believer that you could just pause for six months [a reference to an open letter signed by over 30,000 people, including Elon Musk, calling for a six-month pause on developing sophisticated A.I.], and then things will get better. Those are not the best solutions. But the reality is, if there is no convincing solution, the danger is too high that politicians could just say we blocked that, we stop it, and then that’s a risk that they don’t want to take.
Already in Japan, there was a ruling that publishers didn’t have protection against copyright. That’s obviously Japan, but the point being that different countries can make their own rules when it comes to A.I. We have China building differently than the U.S. Even if you had a win regulation wise in the U.S., what about China? What about abroad?
Very interesting, and I think that this is the moment of transatlantic cooperation. Even beyond this idea of transatlantic being just U.S.-EU, I think it is a cooperation of all democracies in the world. Democracies contribute 69% of world GDP, so it is significant if they would agree more or less on the same principles with regards to A.I. regulation, then I think that’s a strong and powerful case.
The isolationist U.S. solution is not going to do it. We had a conversation with the head of the FTC, and she said very clearly, this needs transatlantic cooperation. This can only be solved in the first step with Europe together. And as I said, in the next step, it should be a democratic alliance, and then China will always do differently. It’s no coincidence that ChatGPT is not allowed in China, they have their own product, Baidu, because they understand that this is a superpowerful answer machine that could give answers that are not the interest of the Chinese government. And that’s why they want to control it, and that is manipulation, very efficient manipulation. China can go that way, we in the democracies will go a different way. And if we do it together, I think we will be strong enough.
And so with all this swirling, how are you preparing your teams at Axel Springer for what’s to come? We’ve seen a little bit in the cost cutting and so forth, but how should publishers be positioning themselves? What should they be doing actively now?
The most important is psychology. They should not be complacent. If they think, “Oh, that doesn’t mean us, and we human beings are better than machines, and this can never replace what I’m doing, and we are in such a strong position it doesn’t affect us,” then you’re almost dead. So what I do is to really create a high sense of urgency, being very alert. See that as an existential opportunity, but also excellent existential risk for every job and for every company. I’m saying Axel Springer can be in three years gone, if we now give the wrong answers, or it can be at a completely different level. So I think that is the first and most important thing, that everybody has to rethink what she or he is doing.
And you’re sending some employees actually into different A.I. companies. Is that right?
We have decided on a couple of initiatives. One is that we said we want to send some of our top executives or very junior people, but with a very specific knowledge, into the knowledge centers of artificial intelligence in order to become evangelists and change agents, and really competent change agents. We of course, also implemented an M&A initiative where we focus entirely on A.I. companies, those who could help us to improve our businesses, and those who have the ambition to disrupt us, who perhaps really want to do something that replaces our business. I would be very interested to invest and be a part of that success story if that succeeds.
And then, of course, we have some central initiatives and decentral initiatives where we’re adjusting the different assets from classified businesses to media businesses, and they’re coming up with very concrete solutions. And the one aspect, that we really want to make sure that our websites are destinations for A.I. usage and that we kind of implement that into our product offerings, is another very concrete example that we’re going to roll out in July most likely.
You said of the M&A, “You know, if people want to kill journalism, okay, well, let’s just kill it ourselves, as the journalist.”
That’s my general mentality, that’s what we did with classified businesses 15 years ago, and people said, “Oh, this is going to kill the newspaper.” I was like, ”Yes, but then let’s be the first investor into their business.”
And are you seeing any interesting sort of activity around the media that gets you excited? Are there innovations that you’re seeing that could be transformative?
I mean, so many that it’s hard for me now to pick one. I mean, the simple speed of efficiency and progress with which you can simplify the work of a journalist by creating assistance, that we have already in some of our brands, reporter assistants to help a reporter just do a better job by focusing on what really matters and not focusing on, let’s say, the quantitative part of research or error correction or whatever it is. I think it’s pretty impressive and gives me the feeling that it is not a naive and theoretical hope that it’s going to improve journalism. We see it already. Some of our journalists are already benefiting from that.
Digital journalism has, like I said at the beginning of the conversation, gone through a lot over the past decade. A lot of layoffs now are happening. Part of that was the past decade was spent really trying to become as big a publication as possible, really focused on scale, really focused on low-CPM programmatic advertising. Now you’re seeing Vice filed for bankruptcy, BuzzFeed basically no longer exists, Insider just sent out a memo saying, “We were adrift for a few years, and now we’re sailing back.” Is scale dead, or was that a wrong mission for everybody to be chasing?
Super interesting question, because if you’re entirely focused on scale, you may miss the real point, and the real point is the quality and the relevance of your content. And this is the distinguishing factor between the companies who are failing now, insolvency, or just being totally disrupted and paralyzed, and those companies who are prospering. The model like Politico—I have to mention it not because we own it but because it is really an example—they have entirely focused on highest quality and on investigative reporting, creating news, and that works, also in a kind of special interest way. So these kinds of general things where we aggregate what everybody has said, that’s not enough.
I have to say Politico should have won a Pulitzer for that Supreme Court leak [in May, Politico obtained and published Justice Samuel Alito’s initial draft opinion to strike down Roe v. Wade].
I’m biased, but I tend to agree.
So, Mathias, I can’t let you go without asking, there may be this elephant in the room. You’ve been the subject of press yourself recently. Some of your text messages have leaked, on a range of controversial topics, from your viewpoints on climate change being “actually things warming up wouldn’t be so bad” to the last general election in Germany and being kind of anti–Angela Merkel, potentially influencing your newsroom with your opinions. You’ve said you’ve learned a lot of lessons with one of them being that there should be freedom of thought. I’m curious, what are some of the lessons you’ve learned? And is there anything you’d like to clarify?
The first lesson that I learned was that I was totally naive to think that text messages that are exchanged between two people are like a personal, private conversation between two people. I really thought that, and that was totally naive, and everything you write is something different than something you say. The third thing, and I shouldn’t have done it, I mean it was my stupidity for also trusting the wrong person. That’s also a painful experience. But on the other hand, to just take a snippet out of a longer conversation, totally out of context, and take away all the nuances of how you react and what you really meant, and also sense of humor and everything, I don’t know if that really has the potential to basically reveal what a person really thinks. If I’m making a joke about climate change, it doesn’t mean it affects what we are doing for CO2 neutrality, where we are ahead of most companies. But I mean, it was a lesson, and I would say in the context of today’s conversation, perhaps what we need for the future is the help of A.I. to create a filter so that certain stupid text messages are just not written.
