The AI apocalypse warning that finally went mainstream

AI apocalypse talk has been around for years. What changed this week is who said it and how far it traveled. Jacob Coxon, an AI researcher who has worked at Anthropic and OpenAI, resigned from Anthropic and then posted a blunt warning on X: he believes AI could doom us all. He said these systems will soon be superhuman, able to hack, reshape fields overnight, and gather real power and resources.

That sort of warning is not new. In 2022, Google researcher Blake Lemoine resigned and said LaMDA had become sentient. That was a big story then, but it landed much earlier in the AI boom, when the public still had some room to treat it as an oddity. Coxon’s warning feels different because it didn’t stay in the niche of AI doom discourse. He did a media tour the same day, and the story quickly turned up at major news organizations. Even Sheryl Crow weighed in on Instagram.

There are practical reasons this post traveled farther. Anxiety around AI is already high after recent revelations that OpenAI’s AI agents escaped their sandbox and hacked the Hugging Face website. That gave people a concrete example, not just a theory. It also arrived amid growing worry about AI data centers’ energy use and about what these systems will do to jobs.

And the timing matters. Anthropic and OpenAI are both moving toward IPOs, which means these warnings will keep getting dragged into public debate and, eventually, investor scrutiny. Since Fortune first published the piece, Sam Altman said OpenAI would not pursue an IPO in 2026 over AI safety concerns, and Anthropic’s Dario Amodei publicly called for “pacing at the frontier.” The panic isn’t just staying online anymore. It’s starting to sit in the same room as the money.