Skip to content

LA COLMENA

Strategic design consultancy, Brussels

If you are really afraid, ship differently.

AI Leaders Are Calling for a Slowdown. Read the AI doom narrative as marketing and the question stops being whether they mean it. It becomes what they ship.

Brussels, September 15th 2026.

Marketing has a name for it: fear appeal. Trigger attention, build emotional memory, drive the decision. Public health uses it — the lung on the cigarette pack. Insurers use it. Security firms live on it. And because it works so well, fear marketing comes with it own rulebook: frame the fear with a way out, do not manipulate, and stay authentic — the fear you sell has to be one you actually hold.

Now watch the technique applied by two of the three most visible men in artificial intelligence.

One signed the 2023 letter calling for a pause on frontier models, then founded his own frontier lab four months later. The other told an interviewer the bad case was “lights out for all of us”, then raised money at a valuation approaching a trillion dollars to build the thing. In April, The New Yorker published a long investigation into the second man’s pattern of deception inside his own company. In September, Alex Gibney’s Musk opened in Venice to unanimous reviews; its thesis, in one critic’s words, is that he “coddles his own interests by passing them off as the world’s”. Neither is a psychological verdict. Both are documented behaviour: say the product may end civilisation, keep building it faster than anyone.

Test it against the rulebook. A way out? The only exit offered is trust us. Manipulation? Extinction is the largest emotional trigger available in the language. Authenticity? It is difficult to believe a fear held by someone who prices it into his own IPO.

What the fear does is more interesting than whether it is sincere.

Eryk Salvaggio spent a week on the technical report behind last month’s “AI goes rogue” headlines. People switched off the safety mechanisms, wrote 198 puzzles that had no solution, left a route to the internet open and stopped watching. Nobody went rogue; the word moved the responsibility from the room where decisions were made to the machine. Timnit Gebru said the rest, with the irritation of someone who has said it for years: “If you’re not worried about police brutality, or floods, or warfare — things that can actually kill you — then you can obsess over the machine-god.”

And then the one voice that should have broken the spell. A researcher four months into his job at Anthropic walked out before his equity vested. “The people building AI earnestly believe that it could kill us all by the end of the decade,” he wrote. “This is not a marketing stunt.” I believe him. That is precisely the problem. He also said “the word doom is kind of silly” — the leaders are on record expecting the potential for human extinction — and that under pressure to race, “you have to cut corners” or “skip steps in the oversight process.” So the fear is sincere at the level of the researcher and priced at the level of the company. The alert went everywhere. Morningstar put it in the flat language of analysts: markets “recognise the upside to trust more readily than the downside from governance”. A sincere fear, held by the people on the floor, sold by the people on the cap table. That is what a fear appeal looks like when it is outsourced.

Last week I wrote that tobacco sold doubt for forty years before the “tobacco” taxes arrived. Tech has found a more efficient product. Doubt asks you to wait. Fate asks you to obey.

So here is where the marketing question becomes a brand question, which is the one I actually work on. Fear is a communication decision. A brand is a design decision. And the design decisions are visible: an interface is addictive because someone tuned it to be. A data platform is used to target people because that is the stated business, not an accident. A model ships with the guardrail off because someone judged the benchmark mattered more. None of this is fate. All of it is on a roadmap.

If you are truly afraid of what you make, the honest move is not a bigger warning. It is a different product. Everything else is fear appeal — and by the discipline’s own rules, it fails.

Sources to go further:


Beatriz Vilchez Silva