AI is fast becoming the way many people encounter the news – through summaries, overviews and assistants – just as traditional readership and ad revenue are under strain. That creates a tension: AI can widen the audience for serious reporting, but only if there’s a sustainable supply of journalism for it to surface.

Last week my LinkedIn feed was a useful case study in how AI and journalism are being pulled in different directions – three threads that together reveal structural pressures on newsrooms and the policy coverage they produce.

1. A newsroom with no journalists

First, my feed carried lively debate among journalists and PRs over the arrival of AgentCrunch.ai: a fully autonomous tech news operation where AI handles topic selection, writing, editing, fact-checking, design and publishing.

Perhaps most surprising, it has no human editors or reporters: a striking experiment in scaling content without editorial headcount. But it raises questions too: about misinformation risks, press releases flowing through platforms without scrutiny, and about what ‘coverage’ actually means when there’s no human judgement involved.

2. A briefing for what’s missing

Secondly, Tom Bristow has launched The Morning Intelligence, a new daily briefing on the UK’s AI ecosystem. His diagnosis rings true: UK coverage oscillates between catastrophe (job losses, existential risks) and euphoria (trillion-dollar valuations), leaving little room for the less dramatic but more important middle ground.

That middle ground includes governance realities – data centre energy demands, sovereignty, and the real trade-offs regulators face between enabling innovation and protecting the public interest. In other words, substance rather than spectacle – and it’s chronically underreported.

3. Publishers pushing back

Towards the end of last week the BBC, Financial Times, Guardian, Sky News and Telegraph publishers launched SPUR: a new industry coalition formed to address unauthorised use of journalistic content by generative AI models. It’s a clear signal that newsroom economics – already strained by audience fragmentation and ad revenue decline – now face additional sustainability questions.

Why this matters for policy comms

These developments coincide with newsrooms at an inflection point. IPPR research argues AI tools are already the “front door” to news for millions, at a time when traditional news readership is dwindling. For many people who would not actively seek out articles or subscribe to outlets, AI summaries and overviews are becoming the way they encounter news at all.

Used well, that’s an opportunity. If AI can handle routine tasks and surface trusted reporting inside the interfaces people already use, it can extend the reach of serious policy journalism to audiences who would never have found it otherwise: by creating new revenue streams, by supporting local and public interest news, and by making trusted reporting easier to find inside the interfaces people already use.

But that only works if there is serious policy journalism to surface. Public understanding of AI’s implications depends on newsroom capacity not just to report competition policy, infrastructure planning and regulatory design, but to challenge them too. The “middle ground” The Morning Intelligence occupies is a reminder of what’s at stake more broadly: treating AI as political substance, and asking whether data centre approvals serve energy security or hyperscaler expansion, whether compute access reaches startups beyond London, whether “competition” claims match real procurement patterns.

AI: shortcut or scrutiny?

These three developments pull in different directions but point to the same pressure: it is hard to fund the kind of journalism that can scrutinise AI governance. AgentCrunch shows what happens when cost-cutting goes all the way: content at scale, but without human judgement. SPUR shows publishers fighting to stay viable. The Morning Intelligence exists because that viability matters.

AI should be taking on routine work and helping people find high-quality reporting. It could free up scarce human time for the harder work of policy analysis, source cultivation and sustained scrutiny that algorithms cannot do. The opportunity here is for AI to be an aid to good journalism, not a substitute for it.