At the start of the month, in a broader announcement about its latest AI features, Microsoft quietly unveiled Copilot Daily which “helps you kick off your morning with a summary of news and weather, all read in your favourite Copilot Voice”.
It goes on to explain how “Copilot Daily will only pull from authorised content sources… such as Reuters, Axel Springer, Hearst Magazines, USA Today Network and Financial Times“. Publishers are being paid – although on what terms we do not know – for content when it is used in Copilot Daily.
This low-key announcement represents a real moment of significance for the news industry. For some time it has been clear that generative AI can be used to create highly-personalised information services; the technology is really good at selecting and synthesising content from a large dataset, based on a set of parameters. But this is the first time these capabilities have been deployed in a news context by a major AI developer, with financials attached for the content creators.
This matters for three reasons. Firstly, services like this give rise to a new set of intermediation risks. Secondly, these risks bring to the fore licensing decisions for publishers which are strategically consequential in the era of AI disruption. Finally, it adds to the pressure on Google’s faltering relationship with publishers, particularly around AI Overviews. Let’s examine each of these in turn.
Related
To date, AI intermediation has been manifest in the form of disruption to referrals from search. This has arisen from two mechanisms:
Firstly, Google incorporating AI Overviews into its search product, thus reducing or entirely eliminating the need for users to visit a ‘destination’ site.
Secondly, via consumers adopting AI tools in place of search. In a new survey The Information has found that more than three-quarters (77%) of its readers are using generative AI tools in place of search and over a quarter report to be doing so in a majority of cases. Whilst this is not a representative sample, this tech-native audience is highly-likely to be a leading indicator of future consumer behaviour. The underlying reason is simple: these tools are just better for certain information retrieval tasks.
[Read more: AI revolution for news publishers is only getting started]
As a result of these mechanisms publishers should expect a structurally-declining flow of referrals from Google. But with Copilot Daily, Microsoft is creating a product which, in some circumstances and for some users, will disrupt an entirely different kind of traffic: that which arrives directly.
Copilot Daily appears to be fairly rudimentary at the moment. But imagine how powerful it could be if it understood what was in your diary that day, in your inbox, on your to-do list, who your favourite columnists are, which media outlets you subscribe to and what specific news stories you’re interested in.
Instead of waking up and checking the FT or New York Times homepage or app, I can bark a command and Copilot Daily – or an AI-powered Apple News product – will give me a highly personalised briefing for my day.
Now clearly this isn’t black and white: even if all those factors fed into the selection of content, many users will still use their news apps and publisher homepages as consumers have always seen value in the editor’s curated view of what is important. But, unlike the effects from AI Overviews and use of e.g. ChatGPT or Perplexity instead of Google which are more likely to impact search referrals, these engagement losses will fall on direct traffic to publisher properties, the growth of which has been a key strategic priority and size of which is perceived to be a crucial measure of resilience.
As a consequence of this development, publishers need to consider a new set of strategic trade-offs. It seems inevitable that, in future, a greater proportion of news publisher revenue will need to come from licensing content as an input to a user-facing service. But in the context of that inevitability – underpinned by the advent of this new technology and its utility in this setting – what does a good deal look like today? And how should any outlet balance the brand and reach upsides with the substitutional downsides.
Beneath these broad questions, sit some granular and fiddly ones. For example, what content can be summarised – everything or just a subset? How long can those summaries be? Should access be provided in real-time or should there be a delay? Should we push to insert a termination clause? Or category exclusivity? How do we want our brands to be represented? Crucially, what is a fair price?
It’s very hard to get these right now, but publishers should be thinking about them and playing forward the likely destination for this technology. Finding the right balance between a loyal audience monetised through engagement on owned-and-operated platforms and a peripheral audience monetised by licensing content to an intermediary service-provider, will be, in my view, the central strategic challenge of the AI era of news publishing.
Finally, this is bad news for Google. Regulators – and publishers themselves (particularly their legal teams) – will all be asking why, if Microsoft is paying to summarise content, shouldn’t Google be too? The only answer being that, thanks to the monopoly position it holds in online search and the consequent imbalance in bargaining power, publishers cannot demand and secure payment.
Regulatory enforcement and judicial proceedings do not move quickly. But over the medium term it’s looking increasingly hard for the Mountain View giant to maintain the position that it will not pay for the use of content to inform AI Overviews. Possibly even general search.
As we see AI deployed in these kinds of settings – and licensing markets emerge to facilitate them – more profound questions about the future of search follow: Will conventional, general search become focused on commercial queries (where the value exchange is clearer) and a new layer of services emerge, built on licensed content, as the main access point for news and broader informational queries?
Predicting the future is hard (and fraught with the capacity to deliver embarrassment) but this certainly feels like the direction of travel.
Email pged@pressgazette.co.uk to point out mistakes, provide story tips or send in a letter for publication on our "Letters Page" blog
