The canary sings again – WAN-IFRA
By Selma Stern
Twenty years ago, the media industry was the first to get flattened by the internet. Print advertising collapsed. Distribution went from a physical monopoly to an open commodity. The audience relationship, once mediated by nothing but the mailbox, was suddenly owned by platforms that publishers didn’t control.
The rest of the economy watched from a comfortable distance. Retailers added e-commerce. Banks added apps. But the core product stayed physical. For most industries, digital was an overlay — a new channel, not a new reality. Media was different because its product was made of the same stuff as the internet: information. When information goes digital, the entire value chain goes with it.
That pattern is about to repeat. And once again, nobody is paying attention to the canary.
The product is the problem
AI is touching every industry. But the depth of that touch varies enormously — and the variance matters.
When a manufacturer adopts AI, it optimises supply chains, predictive maintenance, quality control. The factory still makes the thing. When a hospital deploys AI, it assists with diagnostics, scheduling, drug interactions. The surgeon still operates. The product retains a physical layer that AI cannot replace.
Media has no such buffer. The product — written analysis, reported stories, curated information — is the thing AI produces natively and at scale. One AI tool is already generating over a thousand articles per month for regional publishers, cutting manual effort by 80 percent. Search referral traffic — the lifeblood of advertising-supported publishing — is expected to drop 40 percent within three years.
The disruption here goes beyond workflow. AI is replacing the product itself. And media is the canary again because of a combination that no other industry shares.
First, there’s no physical layer to slow the transition. Second, the relational scaffolding between producer and consumer is thinner than in other knowledge industries. Media does have relationships with audiences — the NYT’s 12.8 million subscribers aren’t anonymous traffic. But compared to a lawyer who knows your case or a consultant embedded in your organisation, even a strong media brand’s relationship with its reader is relatively shallow. And for the reach-driven publishers that spent decades optimising for volume over loyalty, the relationship is thinnest of all. That’s now coming back to bite them.
Third — and this may be the most decisive factor — the consumer replacement cost is nearly zero. Anyone can ask an AI for a news digest right now. Almost nobody can vibe-code their own Salesforce. Software is also a knowledge product under pressure, but the barrier to replacing enterprise software with something you built yourself remains high. The barrier to replacing your morning news consumption with a chatbot is essentially nonexistent.
Media is the one industry where the product has no physical mass, the relational layer is thin, and the audience can replace it with minimal effort. That’s why the canary sings here first.
The scaffolding problem
So what can media organisations actually defend?
Direct access to sources. Proprietary information. Taste and editorial voice. Trust built over years, sometimes decades. These are the things AI cannot replicate — the things that make me renew my NYT subscription every year despite having access to every AI summary tool on the market. I trust them, and I love the writing. The anaphora-laden AI prose flooding the internet makes me want to cry sometimes.
But delivering those things requires being extraordinarily close to the market. Direct audience relationships require product, data, and editorial teams working as one organism. Proprietary information requires authority pushed to the edges — reporters empowered to act fast, close to the ground. Taste requires strong editorial voices with room to develop, which gets diluted by every layer of committee-driven approval. Trust requires speed, accuracy, and accountability. The publishers who are growing — the NYT, the Economist, Le Monde — have built exactly this kind of integration, however imperfectly.
Most media organisations haven’t. They’re still running siloed hierarchies that produce commodity content through slow, fragmented processes — and commodity content is precisely what AI replaces first.
Why? Because the org chart was originally scaffolding for information flow. That was its function. Layers of management existed to route context up and down a chain so that decisions could be made with adequate knowledge. Jack Dorsey said it plainly on a recent podcast: the hierarchy is an old system distributing information across large groups at human scale.
When information starts flowing differently — through AI systems, shared dashboards, ambient data — the scaffolding persists. The information has moved on, but the hierarchy remains, now routing politics, status signalling, and busywork. Structures don’t voluntarily dissolve; they find new reasons to exist. The result is an extraordinarily expensive apparatus doing something nobody asked it to do.
In media, I’ve watched this up close. Editorial hierarchies designed for print production cycles running digital newsrooms. Commercial teams split between advertising and subscriptions as if data doesn’t fuel both. Technology functions — now the most asset-rich part of the organization — reporting three levels below the CEO. The old scaffolding stands, but the information it was built for has found other paths. And the product suffered — because the organisation was optimised for internal coordination, not for closeness to the audience.
Enter the Change Tax I’ve been writing about: the cost of bridging structural gaps with personal effort, relationship management, and sheer willpower. It consists of never-ending alignment meetings and people burning out because they’re compensating for an organisational design that hasn’t been updated since the product was made of paper. That tax was always expensive. With AI compressing every cycle, it’s becoming existential — because the organisations paying the highest Change Tax are producing exactly the kind of generic, undifferentiated content that AI makes worthless.
The convergence nobody coordinated
What’s striking is how many people are arriving at this diagnosis simultaneously — from entirely different starting points.
A few days ago, Ladina Heimgartner, President of WAN-IFRA and CEO of Ringier Media Switzerland, published an essay arguing that media companies’ biggest cost trap is their own structure. AI won’t fix broken models, she wrote. It will expose them. Automating an inefficient process doesn’t create value — it scales whatever already exists, value or irrelevance. Her closing question was sharp: can media companies survive their own operating model?
Dorsey rebuilt Block around an AI “intelligence layer” that gives anyone in the company the same access to information that used to require layers of management to relay. His framing was explicitly structural: the hierarchy’s original function is now obsolete.
Former Navy SEAL Rich Diviney introduced a concept called “dynamic subordination.” Instead of drawing a pyramid or an inverted pyramid to describe high-performing teams, he drew a blob. The leader, he said, is wherever the leader needs to be in the moment. Whoever is closest to the problem and most competent steps up. Jamie Dimon’s annual shareholder letter, published last week, used the same language: teams should be small and authorised to move like SEALs or Delta Force.
A fintech CEO, a media industry president, a Navy SEAL, and the head of JPMorgan — different industries, same conclusion: the hierarchy is the problem. What Heimgartner calls the operating model, what Dorsey calls the intelligence layer, what Diviney calls dynamic subordination — these are different names for the same structural insight. Authority needs to flow to the person closest to the problem. Information can no longer be an excuse for hierarchy.
Nobody coordinated this. That’s the point. When this many people name the same thing independently, the thing’s time has come.
The reason nobody acts
The diagnosis is clear. The convergence is unmistakable. The survival strategy — get close to the market, flatten the silos, stop producing commodity content — is well understood. So why is the structural work so rare?
Because the scaffolding routes something far more personal than politics. It routes identity.
If you built your career climbing a hierarchy — accumulating authority, hoarding information asymmetry, enjoying the view from the corner office — then dismantling that hierarchy requires something far deeper than strategic conviction. It requires giving up the thing that tells you you matter.
Dorsey talks about meditation as a practice for reducing the instinct to react, about deliberately losing credibility with his board and being at peace with it. Nadella replaced Microsoft’s “know-it-all” culture with a “learn-it-all” culture — borrowing the framework from Carol Dweck, a psychologist. These leaders are emotionally equipped to sit at the centre of a network without needing the pyramid to confirm their importance.
Most CEOs aren’t. And in the industry I know best — media — this is where the dysfunction concentrates. I’ve sat in enough C-suite meetings to see the pattern: leaders who understand their structure problem, but cannot bring themselves to act on it, because acting means flattening the very architecture that makes them feel powerful. The hierarchy becomes a mirror. Busting it means breaking the mirror.
This is the hardest thing about the hard thing, perhaps. Ben Horowitz wrote a famous book about how difficult it is to run a company. I think there’s a chapter missing: the inner work that has to precede the structural work. You cannot rewire an organisation if you are the reason it’s wired the way it is. And for a certain kind of leader — one who needs the hierarchy more than the hierarchy needs them — the org chart is important in ways that have nothing to do with the business.
What the canary is telling you
Media is showing every other industry what happens when AI can generate your product, and your audience can replace you with minimal effort. The organisations in the middle — too large to move at speed, too small to absorb platform dependency — are the most exposed. The market is splitting into premium brands that compete on trust and deep audience relationships, and commodity infrastructure at massive scale. The middle is hollowing out.
The same split is coming wherever the relational layer is thin and the consumer replacement cost is low. Education is close, so are parts of financial services or legal services. Each industry’s timeline depends on how thick its relational scaffolding is and how hard its product is to replicate with a prompt. Media’s scaffolding is thinnest and its product is easiest to replicate. That’s why it goes first. Again.
The leaders who are acting now are doing it from a position of choice. The ones who wait will do it from a position of crisis. And the ones who mistake an AI copilot for a structural rewiring — who automate the existing scaffolding instead of questioning why it’s still standing — will scale their dysfunction faster than they ever thought possible.
The canary is singing again. Same song, different verse.
This article was originally published on Selma Stern’s newsletter A Stern Look. It is republished here with permission.
About the author: Selma Stern is a strategy consultant for media & tech execs. Her publication, A Stern Look, examines why organizations lose leverage — and how to get it back. Dissecting the ‘Change Tax’, structure, and power in the post-digital company.
