In a 45-minute CBS "Sunday Morning" interview aired on September 20, Jensen Huang pushed his critique of the AI deceleration camp to its furthest point: "the probability of AI wiping out humanity by 2030 is zero," and "no matter what others do, we should move as fast as possible." Then he went further. The frontier lab leaders warning of AI danger are "dishonest," he said — the public should "read between the lines": what those CEOs actually want is not new laws but relief from existing ones.
[1][2]That accusation is worth unpacking, because it flips the entire slowdown debate. The Amodeis say in public: AI is moving too fast, we need new regulation. Huang's reading: these CEOs talk about new regulation but actually want old constraints gone — if AI gets legislatively classified as an "uncontrolled risk," labs gain a shield: liability relief, subsidies, a way to exclude open-source rivals. His implication: the more alarming the warning, the more specific the interest behind it.
Line up all of Huang's statements in this interview and you get a coherent position, not a soundbite. First, his rejection of doomsday talk is cognitive, not tactical — he does not see AI losing control as a real 2030 problem; "zero percent" is a judgment, not a metaphor. Second, he frames the industry stage as "transitioning from research labs to product manufacturing and services"; in that frame, slowing down is not prudence but refusal to ship — a drugmaker halting after successful Phase III to "study the side effects." Third, he opposes "blindly adding new AI regulations," but note: he opposes the blind addition, not regulation as such. His Goldman line — "the more problems AI creates, the bigger the AI security market" — shows he does not reject regulatory dividends; he rejects regulation in the name of slowing down.
What is most worth recording in this exchange is the shift in narrative authority. Two weeks ago Amodei's slowdown manifesto set the agenda and Huang was defending. A week ago the four companies were sued over the slowdown calls (Buist v. Anthropic PBC), turning "slow down" from moral initiative into legal risk. Today Huang, with one "read between the lines" counter-charge, puts the deceleration camp back under scrutiny: is your call for the public, or for yourselves?
This is not a "who to believe" question; both sides have their own interest spectra and their own sincerity. But one verifiable fact deserves to be pinned on the table: if Huang's reading is right — that the labs' real demand is exemption and deregulation — then in the coming legislative fights you will see them vote on specific liability clauses rather than take positions on "regulation in principle." The real signal is not in the interview. It is in the statutory text.
[1][2]