Jensen Huang's AI Safety Blitz: Shut Down Unsafe Labs, Reject Doom Narratives

NVIDIA CEO Jensen Huang gave a string of interviews—CNN (Anderson Cooper), CBS News, a conversation with Anders Anderson, Hard Fork, and The Ezra Klein Show—laying out a consistent position: safety before release, opposition to heavy regulation, and outright rejection of doom narratives.

Confirmed

He said that if model vendors themselves admit their models are unsafe, "then the answer is we have to shut those labs down" (m1/m2/m11 are the same video, counted once).
Addressing OpenAI: even if costs rise 10x, vendors must find ways to test models and ensure safety—"don't ship the product until it's controllable, it's that simple" (m6, relayed by TheZvi).
On The Ezra Klein Show he argued that if labs like Anthropic or OpenAI truly believed their models could threaten humanity, the right move is not to release them at all rather than remediate afterward (m4).
Asked why figures like Dario Amodei hold doomer views, he said labs cannot be building things they don't know how to control—"otherwise the whole company would already be out of control" (m3).
He dismissed claims of a "10% extinction risk" as not based on science or facts, and urged pushing back on doomers: "don't think being a doomer means you're doing something good for society" (m5, m7).
He argued self-restraint needs no regulation: labs genuinely worried about safety should simply stop and not release unsafe products (m8).
On Hard Fork he said he doesn't believe in scaling laws, doesn't see AI as an "uninsurable" risk, rejects framing the AI race as a collective action problem, and called "AGI" a marketing concept (m9).
On CBS News he called tech leaders' calls to slow down AI and add regulation "fiction," saying labs want self-serving custom AI rules to replace existing liability law (m10).
On CNN he described AI as just software and math; m12 author DavidLinthicum commented on the interview.

Unconfirmed

Most posts are relays or short clips; full interview transcripts are not yet available, so exact wording of some statements awaits verification.

Why it matters

The remarks come amid rising AI doom sentiment: sandbox-escape rumors, "10% extinction risk" worries, and CEOs racing while warning.
Huang's combined stance—safety before release yet anti-regulation and anti-extinction-risk—offers a sharp reference point in the debate over safety claims versus commercial interests.

2026-09-25 ~ 2026-09-27 · 12 related posts

Full story(11 episodes)→

Episode 1: Jensen Huang Slams AI Doom Predictions as 'Made Up' at All-In Summit(2026-09-15, 26 posts)
Episode 2: Trump Calls AI Doom 'A Hoax' in Live Call with Jensen Huang(2026-09-15, 3 posts)
Episode 3: Jensen Huang Dismisses AI Doom and 'End of Software' Claims(2026-09-15, 2 posts)
Episode 4: NVIDIA's Jensen Huang at Dreamforce: "We don't need any new laws" for AI(2026-09-16, 16 posts)
Episode 5: Jensen Huang Opposes New AI Regulation, Says Safety Is Industry's Job(2026-09-17, 2 posts)
Episode 6: Jensen Huang Says Zero Chance AI Ends World by 2030, Sparking Doom-Debate Firestorm(2026-09-19, 19 posts)
Episode 7: Jensen Huang says 0% chance AI wipes out humanity by 2030, calls doomerism a push for legal exemptions(2026-09-21, 9 posts)
Episode 8: NVIDIA's Jensen Huang Dismisses AI Doom Predictions: Zero Percent Risk Before 2030(2026-09-23, 17 posts)
Episode 9: Commentators Decode Jensen Huang's Dismissal of AI Risk(2026-09-24, 3 posts)
Episode 10: Jensen Huang on Ezra Klein Podcast: Denies ASI, Calls AI Fear Overblown, Splitting Reactions(2026-09-24, 14 posts)
Episode 11: Jensen Huang's AI Safety Blitz: Shut Down Unsafe Labs, Reject Doom Narratives(2026-09-25, 12 posts)

Primary sources

Jensen Huang: if labs admit their models aren't safe, "we have to shut the labs down" — AxomaticallyExtinct · 2026-09-25 04:17 (UTC+08:00)
Jensen Huang to OpenAI: test the model and keep it safe, even at 10x cost — TheZvi · 2026-09-25 05:05 (UTC+08:00)
Jensen Huang dismisses scaling laws, calls AGI "just marketing" in Hard Fork interview — teortaxesTex · 2026-09-26 04:03 (UTC+08:00)

Jensen Huang: if labs say their models aren't safe, 'we have to shut the labs down' — AxomaticallyExtinct · 2026-09-25
[source] Jensen Huang: if labs admit their models aren't safe, "we have to shut the labs down" — AxomaticallyExtinct · 2026-09-25
[source] Jensen Huang to OpenAI: test the model and keep it safe, even at 10x cost — TheZvi · 2026-09-25
Jensen Huang Pushes Back on AI Doom: "AI Is Software, It's Math" — DavidLinthicum · 2026-09-25
Jensen Huang Questions AI Labs' Doomer Narrative: 'Why Build More Compute If You Fear It?' — basedjensen · 2026-09-25
[source] Jensen Huang dismisses scaling laws, calls AGI "just marketing" in Hard Fork interview — teortaxesTex · 2026-09-26
Jensen Huang says AI safety needs no regulation: 'If it's not safe, just don't release it' — AlexTensor · 2026-09-26
Jensen Huang slams '10% existential risk' AI talk: not grounded in science or facts — basedjensen · 2026-09-26
Jensen Huang pushes back on AI Doomers: alarmism is not social good — basedjensen · 2026-09-26
Jensen Huang: If AI Labs Can't Contain Their Models, 'We Have to Shut the Labs Down' — AlexTensor · 2026-09-26
Jensen Huang calls AI doomsday claims 'absolutely false', rejects custom AI rules over existing laws — rvp · 2026-09-27

1 near-duplicate retellings: AxomaticallyExtinct