When
What happened to AI-run outlets
Hurwitz described several recent experiments in which AI agents were set up to run an entire media outlet, acting as writer, editor, and decision-maker on what to cover. He chose not to name the projects, but he said none lasted more than three or four months. The outlets did not gather enough readers, he said, their distribution was limited, and their story selection did not match what people wanted to read. Whatever the technical thesis behind each experiment, the audience verdict was the one that counted.
His conclusion was about the reader rather than the technology. “People want to consume content from real people,” Hurwitz said. Spiro agreed with the principle and added a refinement from his own routine: audiences want content from real people, but AI can serve as the aggregator that surfaces it. Hurwitz did not dispute the refinement, and the exchange set up the show’s broader discussion of how AI and human editors can each do what they do best.
AI as a personal aggregator
Spiro described the setup he uses. An automation running on his language models sends a tailored briefing to his phone every afternoon, working in effect as a personal newsletter that opens with “Here are some of the top stories that you may have missed.” Each item comes with an explanation of why it matters for his sector, for the country where he lives, and for the stories he is writing. The stories themselves, he said, are hopefully written by humans, and “AI has kind of personalized it for me.”
That use case captures the balance both men described. The reporting and the editorial judgment stay with people, and AI helps readers find what is relevant to them in a crowded news flow. Hurwitz’s own firm applies a similar sensibility to distribution. Omri Hurwitz Media runs a PR practice for more than 60 companies, about 16 of them unicorns, among them Upwind Security, and a media acquisition practice with more than $16 million in direct equity in U.S. outlets. Part of what the firm brings to those outlets is stronger social distribution, including paid ads and Twitter arbitrage.
Information has always been shaped
Hurwitz was optimistic when the discussion moved to how AI systems source information. He noted that the shaping of information is not new. Before language models, he said, the organizations that invested most in search optimization could influence what people saw on Google, and before the internet, gatekeepers such as major newspapers set the terms. “It has always been like that,” Hurwitz said, and he thinks today’s audiences have more choice than any before them. He called it a great time to decide what information to take in, and he was confident that “AI will not interfere with that.”
Spiro noted that the field is still young. He recalled a reflection that people operate in a kind of fog and only see the road they took when they look back, and he suggested that people may someday ask why every platform did not adopt community notes sooner. He welcomed experiments such as the one by Steven Bartlett, whose podcast embeds community notes in the show before it airs and surfaces them as pop-ups on YouTube. For Spiro, accountability delivered in real time and from the grassroots is a promising direction.
AI companies move into media
The conversation also touched on AI companies becoming media owners. For OpenAI, the entry point was TBPN, the daily tech show it acquired in April and which now reports into its chief global affairs officer. In Europe, ETN, a daily show in the same style, gathered a $1.6 million seed round and expanded to five days a week, with Axel Springer, OpenAI, and DeepMind among its investors. On the platform side, Spotify’s reported $200 million licensing deal for The Joe Rogan Experience was renewed in 2024 and 2026, estimated at $250 million each time. Hurwitz disclosed that his company is trying to buy equity in ETN.
Hurwitz suggested separating two questions. One is media acquisition, which he sees as part of a long history of owners and audiences finding each other. The other is how language models weigh and balance their sources, which he called a matter for their engineers. His overall explanation for the investment wave was short: “it’s eyeballs, brand capital, institutional capital.” His example was Alex Cooper’s podcast: once it is bigger than a CNN segment, brands follow, and institutions after them.
What this means for builders of audiences
For writers and hosts, the lesson Hurwitz drew is that the personal voice remains the asset, with AI as a helper. He said creators can earn from brands, from institutions, and directly from subscribers, and Spiro added that a niche audience of the right people can be enough. Spiro is growing The Spiro Circle for techies interested in Israel, cyber, and geopolitics, and he said its value will rest on disclosure, independence, and a willingness to correct mistakes. In an age of automated writing, the people behind a publication are what readers come back for.
This article was published under HackerNoon's Business Blogging program.





