The industry has rightly spent enormous effort discussing AI licensing and ensuring publishers are compensated when their journalism trains or grounds AI systems.

This is a necessary conversation, but perhaps we should be asking an additional question — one that works for new media and for consumers and may be more palatable in the short term than substantial licensing payments.

If consumers increasingly spend their time inside ChatGPT, Gemini, WhatsApp, YouTube, TikTok, and future AI interfaces (as I discuss in a recent blog), why should a publisher subscription only work on the publisher’s own Web site or app?

What this could look like

Imagine consumers being able to authenticate existing subscriptions once and carrying it across the Internet. The subscription lives with the consumer, not the channel.

A Financial Times subscriber could ask ChatGPT a question and receive premium FT reporting. A New York Times subscriber could access subscriber-only analysis inside Gemini. A local newspaper subscriber could receive exclusive reporting through WhatsApp or another messaging platform. The subscription travels with the customer rather than being tied to a destination.

This isn’t primarily about selling more subscriptions inside third-party platforms, although that should also be possible. It’s about recognising and authenticating the subscriber wherever they choose to consume journalism. Maybe it’s one platform, maybe it’s multiple — they get to choose.

Publishers have spent decades trying to bring subscribers back to owned-and-operated products. But if consumer behaviour is shifting towards platform-native experiences, perhaps the relationship — not the destination — is what matters most.

In many ways, this solution could strengthen rather than weaken the direct relationship. Consumers would continue paying publishers directly, while technology platforms become authenticated distribution channels rather than gatekeepers.

Instead of forcing users to choose between the convenience of AI and the value of trusted journalism, publishers could offer both.

Perhaps alongside licensing discussions, the industry should be working with technology companies to develop an open authentication layer that allows publishers to recognise subscribers across AI assistants, messaging platforms, and other digital experiences.

This would benefit everyone: Consumers get seamless access to journalism they already pay for, publishers extend the value of their subscriptions, and technology platforms gain access to trusted, high-quality information without replacing the publisher relationship.

So why hasn’t this happened already?

If this feels like a win for publishers, consumers, and technology platforms, why isn’t anyone building it?

At first glance, it seems like an attractive proposition for the platforms. And in reality, some are at least thinking about it

Earlier this year YouTube said they were, then walked it back (more here). Apple Podcasts already supports publisher-managed subscriptions, while Spotify’s Open Access programme lets participating publishers authenticate existing subscribers to unlock premium podcast content. And startups such as Stardust are offering WhatsApp authenticated for direct chat with audience/subscribers.  

I’ve argued before that AI companies will likely exhaust every possible commercial model before committing to large-scale licensing payments. An authentication layer could become one of those “alternative” models — helping publishers generate direct subscription revenue while reducing some of the pressure around licensing.

Offering authentication could also be commercially attractive in its own right.

News subscribers tend to be more affluent, better educated, and already comfortable paying for digital products. They’re exactly the kind of users technology companies want to retain — more likely to upgrade to premium AI models, more valuable if AI shifts towards commerce and affiliate revenue, and ultimately more attractive to advertisers.

So what’s the catch?

The challenge isn’t whether it’s possible. It’s whether it’s worth prioritising.

Every product manager knows the first release is the easy part. The hard part is everything that comes afterwards.

Publishers won’t just want subscriber authentication. We’ll want frictionless subscription sales. We’ll want analytics dashboards. We’ll want customer support workflows, billing options, APIs, security, privacy controls, product improvements, and feature requests. Then we’ll want to test new ideas and ask for even more.

None of that is core to an AI platform’s business today.

If we genuinely believe this is the future, the industry can’t turn up with hundreds of different feature requests and competing priorities. We need to be cohesive, realistic, and incredibly clear about what we’re asking for. Start with the smallest viable product that creates value for everyone, prove it works, and build from there.

The idea makes a lot of sense to me.  

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

Banner photo: Adobe stock image by Sandwish.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .