The conventional wisdom a few years ago was that artificial intelligence would make public relations obsolete. If AI could generate content at scale, summarize information instantly, and surface answers without sending readers to media outlets, the earned media model that PR firms depend on would erode. Omri Hurwitz watched that argument develop, felt the pressure it created, and then watched it collapse.
In a conversation with Yoel Israel, Hurwitz described what actually happened when large language models became mainstream. The full exchange is on YouTube and is worth watching for anyone trying to understand where media credibility and AI distribution intersect.
His account is straightforward. As AI systems began training on internet content and serving answers drawn from that content, the value of being mentioned in credible media outlets increased rather than decreased. LLMs do not generate reputations from nothing. They reflect what exists on the web, weighted by the credibility and reach of the sources where information appears. A company with consistent, high-quality media coverage across authoritative outlets is a company that AI systems learn to describe favorably. A company with no coverage, or coverage only on low-authority sites, is a company that AI systems have little to say about.
“Every mention of you on the internet is PR,” Hurwitz said. “And now think about my firm’s assets and distribution. Who’s better positioned?”
The question was rhetorical. By the time AI became a mainstream business concern, Hurwitz had spent years acquiring media assets, building coverage infrastructure, and developing distribution capabilities that most PR firms did not have. The strategy he had pursued for reasons that had nothing to do with AI turned out to be precisely the right infrastructure for an AI-shaped media environment.
Israel was candid about his own earlier skepticism. He had assumed traditional media would become less significant over time, that audiences would migrate entirely to new media formats and that the authority of established outlets would diminish accordingly. What happened instead was more nuanced. Traditional outlets retained their authority as training data sources for AI systems even as their direct readership numbers plateaued or declined. The outlet you might not read directly is still shaping what an AI tells someone about your company when they ask.
Hurwitz also pushed back on a framing that treats AI and PR as separate tracks. In his view, they are converging. The firms that understand how AI systems source and weight information, and that build their media presence accordingly, will have a structural advantage over firms that are still thinking about PR purely in terms of human readership and click metrics.
He was also clear about what this does not mean. It does not mean gaming AI systems with low-quality placements across high volumes of marginal outlets. The credibility weighting that LLMs apply means that a placement in a genuinely authoritative outlet carries more signal than dozens of placements in outlets with thin audiences and low domain authority. Quality still matters. It may matter more than it did before, precisely because AI systems are making credibility assessments at scale.
For founders and marketing leaders trying to understand where to invest in brand building right now, the argument Hurwitz makes is that the earned media work you do today is training data for tomorrow. The companies that are visible, credible, and well-covered across authoritative sources are the companies that AI systems will describe accurately and favorably when a prospective customer, investor, or partner asks.
That is a different way of thinking about PR than most people in the industry are used to. It is also, based on how the landscape has actually developed, a more accurate one.