The take that PR is dead is wrong, and the take that PR works the way it did in 2020 is also wrong. Both miss what’s actually changed.
Journalists use AI now. In Muck Rack’s 2026 State of Journalism survey, 82% said they use at least one AI tool, and Google Discover picks stories using many of the same automated systems as Search. The earned media surface is still there, it’s just operating at machine speed underneath, and the brands that get that win.
What’s shifted
Pitch volume hit a wall. Most PR agencies now have AI drafting outreach, so journalists’ inboxes are filling with generic, well-written and useless pitches. The same Muck Rack survey found 88% of journalists immediately disregard pitches that miss their beat, so generic outreach is functionally dead. (The earned-media pitch generator playbook in The Lens matches each story to the journalists who cover that beat.)
Quotes get plucked, not requested. Journalists can find usable quotes on the web without waiting for a reply, so if your senior people have visible, sharp opinions in long-form those quotes get picked up, and if they don’t you’re invisible.
LLM-cited matters more than journalist-cited. Being mentioned in a chat answer when someone asks the model about your category is the new version of being in the trade press. The mechanisms are different (answer-engine optimisation matters here) but the value is similar, third-party validation, surface area and brand-search lift.
Social platforms are aggregating taste signals. Algorithmic feeds use AI to predict what to surface, so branded content that doesn’t earn engagement gets buried faster, and the bar for “good enough” rose.
Where earned actually compounds now
Three places are working harder than they used to.
Founder and operator voice. Senior people publishing real opinions in their name, on LinkedIn, in long-form, in podcasts, generate the citations and pull-quotes that get re-used everywhere else. It works when the perspective is specific, opinionated and hard-won, because generic posts don’t pull. (Senior social posts are one of the five places we say AI shouldn’t touch, and the founder and institutional voice playbook sets out who speaks for the brand where.)
Original data. Surveys, benchmarks, weird internal numbers shared publicly. Journalists need the data and LLMs need the source, so both link back, and if you can publish something genuinely novel quarterly you out-earn every press release.
Partnerships with edge. The boring co-marketing brief is dead, but the interesting partnership where two brands do something genuinely new together earns coverage because it’s an actual story.
What stops working
Press releases for the sake of them, round-up “X best of Y” link bait, generic op-eds drafted by an AI and sent through an AI to be edited by an AI.
The honest play
Earned channels still work, and what earns coverage hasn’t changed. Real opinions, real data, real partnerships, distributed at speed, with AI helping you operate the surface around them (research, drafting, monitoring, repurposing) without diluting the substance. (The five-point workflow for that production layer is broken down in five practical ways to improve content production with AI.)
The brands losing earned right now are the ones using AI to automate the wrong layer. They’ve automated the substance and left the workflow human, which is exactly backwards.