AI sparked the third age of publishing: will it empower or destroy?

OpenAI and Axel Springer - owners of Politico, Business Insider and German titles including Bild and Welt - announced ‘a global partnership to strengthen independent journalism in the age of AI’.

Their words, not mine...

The deal will see Axel Springer license content from across its media brands - including that which is paywalled - to the ChatGPT developer. The chatbot’s users will get summaries of articles with citations and links back to Axel Springer sites.

This represents the dawn of a new, third age for digital publishing. In each of the two so far, publishers have taken far too long to adapt their strategy to changes in society, technology and how the former use the latter. I fear the same will happen again.

Here’s my extremely reductive take on the story so far and how this deal fits in.

First age: the side hustle - 1997 to 2010

At the beginning, there was BBC news online… Our august national broadcaster was the first mainstream UK news outlet to open an online presence - in November 1997 - and it typified the approach of quality news outlets. In short, the website was an innovative side project. It wasn’t about generating revenue but experimentation with the format and building an audience as consumers started ‘surfing’.

Fewer than 10% of UK households had an internet connection when it launched. By 2008, over 65% did. The approach of the publishing sector can be summarised as ‘build a product, attract an audience, and the advertising revenue will follow’.

This was inspired by the huge valuations of nascent platforms (in early 2008, Facebook was worth $16bn on the back of a $140m loss the previous year). Paywalls were anathema to the prevailing ‘information wants to be free’ philosophy of the internet. But the ad money didn’t really materialize; publishers hadn’t built products that consumers would pay for or even created a perception that news comes at a price.

Second age: the rise and fall of social news - 2010 to 2023

While search had always been an important traffic source, in the early 2010s, social media rapidly rose to become a crucial news source for many consumers. By 2017, over half of US consumers accessed news via social platforms – twice the number that did so in 2013.

Despite growing online audiences fueled by social, digital advertising revenue simply wasn’t landing for publishers; instead it was accruing to Google and Facebook. And meanwhile the legacy revenue engine of print advertising was in steep decline as media investment followed eyeballs. Publishers were on their knees. Along came the paywall.

The FT was ahead of the pack, launching in 2007. The Times followed in 2010, The New York Times in 2011 and The Telegraph in 2013. It’s been a slow burn, but finally, premium publishers, and particularly those with differentiated content and affluent audiences, have been able to make this model work.

Let’s not overstate it, though. In the UK still only 9% of consumers pay for news. In the US - where the BBC isn’t providing high-quality free online journalism - that number is 21%. These figures are now flatlining, so publishers are fighting over wallet share. And there are strong suggestions that the market - for general interest news at least - will be dominated by a small number of large players that benefit from scale and talent (the New York Times accounts for over 35% of subscriptions in the US). And the model for mass-market and tabloid titles is far more precarious.

Meanwhile, there’s been a decoupling between social media and news. After the controversies of the 2016 US election and Cambridge Analytica, Facebook has doubled down on social content and deprioritized news. According to Meta, it now accounts for less than 3% of what people see in global Facebook feeds3.

Putting aside their continued reliance on search traffic, this is a healthier place for publishers; social-native businesses that relied on click-bait have folded, and those that remain are by and large focused tightly on the user and serving their needs. But it’s taken 25 years to get there.

Third age: AI intermediation - 2024

Just as news publishers find a way to keep the lights on - although perhaps dimmed - in the digital age, along comes the next technological wave - AI.

Chatbots, right now, are really not great at dealing with news queries. But given the inflows of capital and talent, it’s beyond doubt that it will improve. AI executives talk about an ‘agentic future’ in which digital assistants perform all kinds of actions for us. It’s hard to imagine that information and news retrieval aren’t going to be one of those.

For publishers, that means intermediation. As the sector is currently oriented, almost all publisher revenue - whether that’s selling subscriptions, advertising, or event tickets - relies on users visiting and consuming content on publisher-owned platforms.

In this AI-powered future world, it seems inevitable that the industry will need to pivot its model such that a greater proportion - and perhaps even the majority - of revenue is derived from licensing journalism as an input to user-facing services. Just like Axel Springer has done with OpenAI.

So, in one sense, the deal is a portent of the future, and Axel Springer should be praised for getting it across the line. But is the number - reported as being in the tens of millions of Euros - high enough for a business with billions of euros of revenue, given the risk to its existing model? On that, I’m really not sure...

In my next column, I’ll dig a little deeper into what it will mean if ChatGPT becomes our primary news source. God forbid.

David is director of public affairs and platform strategy at the Financial Times. His views stated above are his own, not his employer’s. Continue the conversation with him on LinkedIn.