Cal Newport: AI Labs' Doom Rhetoric Is Morally Indefensible

When a company's loudest message is that its own technology might destroy civilization, the ethical next move is presumably to stop. Cal Newport, a computer science professor at Georgetown, argues in The New York Times that major AI labs are doing neither: they warn loudly and keep building. He calls this "doom trolling," and says he has come to believe it is "one of the defining and most arresting properties of our current A.I. moment" and "morally indefensible."

Newport's argument turns on a binary. If OpenAI and Anthropic genuinely believe their systems pose the catastrophic risks they publicly describe, the only ethical response is to stop. If they don't actually believe the warnings, then the doom rhetoric is something worse: Newport describes it as "laundering the anxiety of millions to improve the financial fortunes of a vanishingly small number of major stockholders." Either way, he argues, the companies should stop "treating A.I. like some inevitable force that they're struggling to steward" and instead explain "the benefits of their tools" and "justify their costs," like any normal product company.

Specific cases drive the piece. Newport cites Anthropic's report on AI coding agents achieving recursive self-improvement, a scenario in which the company warned it could lose control of its own systems, while continuing development regardless. His analogy: no car manufacturer would publish a whitepaper warning its vehicles might spontaneously catch fire and then keep selling them without a remedy.

Newport does credit OpenAI's recent paper "Built to benefit everyone" as a return to conventional, optimistic product language, treating it as evidence that the doom register is a choice, not an inevitability.

The piece's limits are real. Technical risks exist on a spectrum Newport's binary does not fully address, and the essay does not engage with whether even cynical risk warnings can produce useful policy outcomes. The piece drew quick endorsement from Chamath Palihapitiya, who called it "an excellent essay," and from David Sacks. That uptake signals "doom trolling" is already becoming contested political vocabulary, and how the framing settles into AI safety regulation debates is worth watching closely.