AI companies built the danger and published the warning. They kept building. The warning was the brand. The danger was the product.
Literal meaning: Doom trolling is the deliberate strategy by which AI company leaders spread apocalyptic narratives about their own technology — claiming it poses existential risk, threatens human civilisation, or is on the verge of becoming uncontrollable — while continuing to build and fund it. The term was defined and popularised by writer and professor Cal Newport in a 2026 piece for The New York Times: the warnings are not honest risk assessments but calculated provocations that serve the companies spreading them.
Origin: Newport’s diagnosis emerged in response to a wave of public statements by AI executives — predictions about artificial general intelligence arriving within years, about AI systems “going rogue,” about civilisational stakes. What Newport identified was not the content of these warnings but their structural function: each doom statement increased the speaker’s visibility, inflated the perceived importance of their product, and implicitly established that only their company had the expertise to manage the danger. A broader pattern had already been described by Johnson (2022), who called out “doom trolls” as figures exploiting catastrophist discourse for engagement. Newport sharpened this into a specific critique of AI industry behaviour — and gave it a name.
They warn that their product is an existential threat. They are still building it.
The Appeal: The doom narrative is structurally compelling. It sounds responsible: look how seriously we take the risks. It sounds humble: even we are afraid of what we are making. It speaks to science fiction intuitions that audiences already hold — Skynet, the Matrix, Digital Frankenstein. The language of existential risk, drawn from Longtermism and The Singularity discourse, lends the warnings philosophical depth. And because the speaker is positioned as a technical expert, the narrative is received as information rather than rhetoric.
The Friction: This is not borrowed SF — not the Terminator, not Asimov. It is self-generated speculative fiction, authored by the companies building the technology, about their own technology, in their own financial interest. That is what makes it bizarre and what makes it a market mechanism rather than a warning.
The economic logic: existentially dangerous = existentially valuable. A technology that could end civilisation justifies billion-euro investment rounds, demands regulatory frameworks that entrench incumbents, and commands permanent public and political attention. The doom story is the fundraising deck. This is why the warnings never lead to the speakers stopping, pausing, or handing control to genuinely independent regulators: the danger is narrated but never acted on.
The pattern extends beyond any individual statement. When one lab claims existential risk, competitors must respond in the same register or appear naive. Doomerism becomes mimetically contagious — see Mimetic Desire: the doom competition escalates not because the risk is better documented, but because each lab imitates the other’s rhetoric.
Why This Matters: Doom trolling is not Doomerism. Doomerism is a genuine affective and political position — the sincere belief that catastrophe is unavoidable. Doom trolling is a strategy. The line between them is deliberately kept unclear: a CEO can appear to be a doomer while functioning as a doom troll. That ambiguity is the mechanism.
The stakes of the confusion are concrete: doom trolling monopolises attention on speculative future dangers while directing it away from present structural harms — algorithmic bias, surveillance infrastructure, labour displacement, energy consumption. The existential AI that might arrive in ten years displaces the Surveillance Capitalism that is operating right now. Once you can name doom trolling as a strategy, the question changes from is the doom probable? to who needs you to believe it is?
Related terms: Doomerism · Longtermism · The Singularity · AGI · Mimetic Desire · Digital Frankenstein · Surveillance Capitalism · Skynet · AI Literacy
Read more:
- Dear A.I. Companies, the Doom Trolling Has to Stop — Newport, C. (2026, 17 June). The New York Times
- Dear AI Companies: Stop the “Doom Trolling” — Newport, C. (2026, 22 June). calnewport.com; also as Deep Questions podcast episode (25 June)
- “Doom Trolling:” Why AI Leaders Are Stoking Fear About Their Own Tech — Amanpour and Company, PBS (2026, 26 June). Interview with Cal Newport
- The AI Industry Must Stop Doom Trolling w/ Cal Newport — Zitron, E. (2026). Better Offline podcast
- Feeding the Doom Trolls Comes at a Cost — Johnson, E.B. (2022, 14 November). Medium. On “doom trolls” as social-media doom-mongers, before the term’s AI use
- The TESCREAL bundle: Eugenics and the promise of utopia through artificial general intelligence — Gebru, T. & Torres, É.P. (2024). First Monday, 29(4). The ideological bundle behind AGI doom talk, and who benefits from it