The industry is currently obsessed with a survey. Trade publications are asking publishers how they feel about AI search, as if a collective "we’re worried" will somehow stop the tectonic plates from shifting. This is the "lazy consensus" of the media world: the belief that if we just document our anxiety and lobby for a few crumbs of attribution, the 2010s will magically return.
They won’t.
The traditional search-to-publisher pipeline is not just leaking; the pipe has been melted down to build something else entirely. Most editors are busy arguing about whether Google’s AI Overviews should cite them more clearly. They are asking the wrong question. The real problem isn't attribution. It’s disintermediation.
If a machine can synthesize your thirty-page deep dive into a three-sentence bulleted list that satisfies 90% of user intent, your business model wasn't "journalism." It was "friction."
The Attribution Myth
Publishers are begging for links in AI responses. This is a loser’s game.
I’ve spent fifteen years watching media companies chase the dragon of platform traffic. First, it was the Google "Ten Blue Links" era. Then the Facebook "Pivot to Video" disaster. Now, it’s the "AI Citation" era. Each time, publishers act like partners when they are actually just unpaid R&D for the platforms.
Let’s look at the mechanics. In a standard SERP (Search Engine Results Page), a link is a gateway. In an AI-generated answer, a link is a footnote. Nobody clicks footnotes. Data from various click-through rate (CTR) studies suggests that featured snippets already cannibalized a massive chunk of traffic. AI search pushes that to its logical conclusion: zero-click search as the default, not the exception.
If you are "responding" to AI search by optimizing for citations, you are rearranging deck chairs on the Titanic. You are optimizing for a 0.01% click-through rate while the LLM (Large Language Model) consumes your entire corpus to ensure the user never needs to visit your site again.
Stop Protecting the Archive
The standard response is to block crawlers. "Don't let GPT-5 eat my content for free!"
This feels proactive, but it’s actually a slow-motion suicide. If you block the crawlers, you don't stop the AI. You just ensure the AI is trained on your competitor’s data—or worse, on the hallucinated garbage of the open web—while your brand vanishes from the only interface users will use for the next decade.
The nuance the "survey crowd" misses is that scarcity is no longer a moat. In the old world, having the best information made you a destination. In the AI world, information is a commodity. The LLM can find "the facts" anywhere. What it cannot replicate is the perspective, the personality, and the community that exists around the information.
The Death of the Generalist
The biggest victims of AI search won't be the niche specialists. It will be the "generalist" news sites and the "how-to" factories.
If your article is titled "How to Change a Tire" or "10 Best Coffee Makers," you are already dead. You just haven't stopped breathing yet. An AI can synthesize that information better, faster, and without twelve paragraphs of SEO-fluff about the history of the wheel.
I’ve seen legacy publishers pour millions into "service journalism" that is essentially just rewriting product manuals. That entire revenue stream is evaporating.
The only way to survive is to lean into what a machine cannot do:
- Primary Sourcing: Machines can’t pick up a phone and call a disgruntled CEO.
- Physical Presence: Machines can’t tell you what the air felt like at the stadium or how the new iPhone feels in a sweaty palm.
- Taste: A machine can tell you a movie’s plot; it cannot tell you if the movie has "soul."
The New Math of Media
We need to stop talking about "sessions" and "users." Those are ghost metrics in an AI world.
The new metric is Direct-to-Device Equity. If you don't have a direct relationship with your audience—via email, a proprietary app, or a physical subscription—you don't have a business. You have a temporary pass to sit in Google’s waiting room.
Consider the economics of a typical ad-supported site. You need millions of views to make a dent because programmatic ads pay pennies. AI search cuts your traffic by 60%. If your response is to "take a survey" or "write more content," you are failing at basic math.
The contrarian move? Shrink.
- Kill the fluff. Stop writing for the algorithm.
- Fire the SEO consultants who tell you to use "long-tail keywords." The AI knows what the user wants better than your keyword tool does.
- Pay for talent, not volume. One writer with a dedicated following of 10,000 people who will pay $10 a month is worth more than a newsroom of twenty people chasing 10 million fly-by visitors from search.
The Mirage of Licensing Deals
Every major publisher is currently salivating over licensing deals with OpenAI, Google, and Perplexity. They think these checks will replace the lost ad revenue.
They won’t.
These deals are "hush money." They are designed to keep you from suing while the platforms build a replacement for your entire industry. Once the models are sufficiently trained on your archives, the value of your "fresh" daily content drops. The platforms are buying time, not a partnership.
If you take the money, use it to build a platform that doesn't rely on search. If you use it to fund your existing "search-first" newsroom, you are just funding your own funeral.
The Infrastructure Pivot
Instead of asking "How is AI search changing us?", we should be asking "How do we become the API for the AI?"
Imagine a scenario where a high-end financial publication stops trying to get people to visit its homepage. Instead, it builds a proprietary "Expert Layer" that plugs into AI assistants. You don't read an article; you query the publication’s specialized, verified data through a secure interface. You charge the AI companies for every query that hits your proprietary insights.
That is a business model. Trying to get a "Source: The New York Times" link at the bottom of a ChatGPT response is a charity project.
Your Content is a Product, Not a Destination
The era of the "Website" as a destination is ending. We are moving into the era of the "Content Fragment."
Your information will be sliced, diced, and served in bits across a thousand different interfaces—from smart glasses to car dashboards. If your business model depends on a user seeing a banner ad next to a 1,000-word article, you are finished.
The "survey" approach is a symptom of a legacy mindset. It’s the sound of people who are comfortable with the status quo trying to negotiate with a hurricane.
Stop looking for a "response" to AI search. Start looking for a way to exist without it.
Burn the SEO playbook. Fire the people who talk about "optimizing for AI Overviews." Hire people who know how to build a brand that people will type into a browser directly.
If you aren't a destination, you’re just data. And data is free.
Build something that isn't.
