"Not Even True": Fact-checking Buck Sexton's climate aside inside an AI doom episode

Claim: current climate change is not unprecedented, which Buck Sexton called "not even true." Verdict: mostly-false (truth score 0.2, confidence 0.9).

Fact-checked: The Buck Sexton Show, "Buck Brief — Will AI Destroy Humanity Like Skynet in Terminator?" (10 September 2026, at 1:17).

The episode is about AI, not climate. It came out one day after researcher Jacob Coxon resigned from Anthropic and warned that frontier labs are "gambling with our lives." Sexton's argument is an analogy: climate scientists had career incentives to sound alarms, so we should discount AI researchers who sound alarms. The climate denial is the premise the analogy rests on, and it goes by in a seven-word aside that Sexton concedes straight away ("but even if it is, we'll all be okay").

First, the strongest case for Sexton. Earth has been far hotter than today. Judd et al. (2024, Science) put global mean surface temperature at roughly 11–36°C over the last 485 million years, and CO₂ was once well over 1,000 ppm. Judith Curry adds a real technical objection: deep-time proxy records smooth out short events, so a brief, sharp spike far in the past could be invisible to us.

Why it still scores mostly-false: the IPCC never claimed an all-time record. AR6 limits its findings to specific time windows. Warming since 1970 is faster than any 50-year period in at least the last 2,000 years, and CO₂ in 2019 was higher than at any point in at least 2 million years. The proxy records for that 2,000-year window resolve individual years. Neukom et al. (2019, Nature) found that the Medieval Warm Period and Little Ice Age were regional and out of sync, while 20th-century warming is coherent across more than 98% of Earth's surface. On rate, the closest fast-warming event in the geological record, the PETM, released carbon roughly an order of magnitude more slowly than we are. The Judd et al. paper that supplies the kernel of truth also found temperature and CO₂ moving in lock-step, and its lead author has objected publicly to its use as proof there is nothing to worry about.

We also trace where the argument came from, from "the climate has always changed" through the hockey-stick fight to a cropped version of the Judd graph, and how it ended up as a precedent in an AI segment. We also cover what we still don't know. We hold only a short excerpt, and Sexton says "as we should discuss," so a fuller case may exist later in the show. He never names the predictions he says failed. PETM onset estimates carry real error bars.