We are conducting research on publisher adoption of AI in their operations, and how they’re balancing the need to protect credibility with the imperative to stay ahead of a massive technological change. Please take a moment to fill out this short survey
On Wednesday, The Rebooting Morning Salon turned its focus on AI strategies. Executives from leading publishers gathered to discuss what’s working in adopting AI to drive efficiency and create new products. The overall tone was one of pragmatism. In this room, most had arrived at the acceptance stage of Google Zero grieving. An underlying theme was how to reclaim agency – in blocking bad actors, piloting new initiatives and developing internal AI tools – rather than adopt a reactive posture.
We conduct our gatherings under the Chatham House Rule to encourage candor vs PR-crafted talking points. Below are some anonymized insights.
Block first, then negotiate. It turns out that not all bots are created equally. Much of the focus in AI crawling focuses on well-known names, but there are hundreds of bots out there collecting publisher data and effectively reselling it for pennies on the dollar. The solution advocated by one large publisher: block by default, switch to an allow list, play a game of Wack-a-Mole with unscrupulous parasites. The prevailing consensus: Even if you’re getting paid, don’t expect it to last. Even Charlie Brown eventually tired of trying in vain to kick the football.
Key quote: "Fairness is not the world. It's a knife fight out there.”
LLM citations as the new SEO. The double edged sword of chatbots as a new information interface is that they do not send much traffic to publishers but presence in their answers is critical. One niche publisher recounted how their clients are focused on how to pop up in answers and see their programs with the publisher as a key part of the emerging, if murky field of AEO, aka answer-engine optimization. The positive view: AI needs fresh information and original reporting, particularly of non-public information, is valuable.
Key quote: "Our placement in the LLMs has to do with the originality of the content we have — a lot of it baked into the models."
MCP and agentic tools are changing workflows. Model context protocol is a wonky but important shift to workflow by allowing data normally trapped in software platforms and dashboards to be accessible via agentic tools. That’s to say, you can use Claude to set up automations and do analysis through the equivalent of a text message. This points to a flattening of workflow with fewer hops to those with “hands on keyboard.” This will result in a cultural shift. Executives floating at the “strategic level” will need to be more executional. Also: Fewer people will be employed in pure executional roles that revolve around around technical familiarity with tools.
Key quote: "I don't really code anymore. I think of business problems and come up with solutions. All of our coders are like that now — they've turned into product people in the last year."
AI tooling is moving beyond production. News publishers are increasing the use of AI in the reporting process. Much of the resistance to the use of AI assumes it will mean robots creating slop. One common use: Finding signal in the noise of unstructured data. For instance, publishers are using AI to monitor conversations on podcasts. One publisher spoke of “vectorizing” its extensive archive in order to make its large corpus friendly to bots, in preparation for a future when websites are created more for bots than humans. AI is inching into tasks like headline writing. Others are using it to drive deeper engagement on their sites, although the use of tools like AI-enabled site search remains low. Another use is: using AI to drive higher ad performance in promoting stories to drive subscriptions.
Key quote: "Everyone's saying we're gonna end up with loads of crappy articles. We kind of did that when we were all chasing SEO — humans created a lot of slop themselves.”
Cultural change is hard. The biggest lie is that people embrace change. People hate change. AI is a massive change that is very unpopular. Executives often grouse about foot dragging and sometimes outright refusal to use home-grown AI tools, often with news organizations citing trust issues and hallucinations, even if their bosses believe these concerns are, at best, motivated. One piece of advice from a newsroom leader: Show don’t tell. Stop talking like Dario Amodei and the VC class with their sweeping pronouncements. Instead, pilot small projects that tackle specific pain points. Get the quick wins, build momentum, and compound buy-in. One note: Publishers are reticent to track and require AI use among staff outside of development roles. I suspect that will change as companies inevitably become more coercive.
Key quote: "Our jobs are not safe from AI — that's resolved, next question. The question to be asking is, what is my job going to be next? It's actually very exciting if you can let go of that first part."
Thanks to Beehiiv and ProRata for sponsoring the salon.
Learn more about why major publishers like Time, Newsweek and Texas Tribune are using Beehiiv to power their email infrastructure.
Learn more about ProRata’s Gist platform helps publishers benefit from the question economy.
Note: All of TRB’s salons and events are invitation only. Register your interest for an invitation here.