Geoffrey Irving, co-founder and chief scientist at the AI safety nonprofit Resolution, told Reuters that “the risk is ramping up pretty fast,” and pushed back directly on the industry's common framing that slowing frontier AI development is an unsolvable coordination problem requiring every lab to act simultaneously. “They could just stop unilaterally,” he said of individual AI companies.
Irving's point cuts against the most common industry defense for continued rapid development, that any single company pausing just cedes ground to competitors who won't. His counterargument is that this framing is itself a choice labs have made to justify continuing, not an inescapable structural fact about how AI competition actually works.