Jensen Huang tells Ezra Klein that AI is just software whose existential risks are overstated, yet his standards would shut down OpenAI and 10x safety spending
Jensen Huang accidentally called for shutting down OpenAI and intentionally called for spending vastly more on safety.
Don't Worry About the VaseZvi Mowshowitz
Context & Ripple Effects
Huang had already framed AI as a net job creator and pushed back on doomerism in a September Q&A on AI jobs and open models. He also argued that AI leaders seeking regulation may be pursuing relief from existing laws for ulterior reasons, sharpening the significance of a safety agenda built around company conduct rather than new rules.
The interview sets his anti-existential-risk rhetoric against a demanding operational test: safety claims carry consequences only if labs are expected to withhold products that fail them and fund the work needed to meet that bar.
First-order effects
Under Huang’s stated standards, OpenAI would have to stop operating unless it met the safety threshold he describes.
A tenfold increase in AI safety spending would redirect substantially more frontier-lab resources from model development and commercialization toward safety work.
Second-order effects
OpenAI’s peers face a clearer cost-and-credibility comparison: safety commitments become harder to separate from the staffing and budgets required to enforce release decisions.
The disagreement shifts attention from abstract existential-risk claims to who sets product-release standards and bears the cost when a model fails them.
Third-order effects
If leading AI companies adopt Huang’s framing, AI governance may develop through enforceable release discipline and liability exposure rather than the broad new regulation he has criticized.
The resulting divide would be between labs able to absorb large safety budgets and those whose business models depend on faster, cheaper deployment.
The trend: AI-safety debate is moving from arguments over existential risk toward whether frontier labs will fund and enforce operational standards strong enough to constrain releases.
it is completely unsurprising that an actual ML scientist in Dario does not jive well with a world of professional talkers & wheelers/dealers in Huang, Trump, even Altman. Amodei is the only major executive whose published work I have ever found useful in my own job.
The Jensen interview is interesting in that he sounds so much more like an adult in the room than the executives leading the frontier labs. no philosopher king stuff, very practical concerns. also: he seems almost completely unable to think hard about risk lol
🚨 Jensen Huang Says Something Seems REALLY Off About the AI Labs' Doomer Narrative JENSEN: “I really don't believe that they're building something they have no idea how to control. Their whole company is out of control...” ANDERSON: “So why are they saying this?” JENSEN:
I thought Jensen Huang made plenty of good points with @ezraklein. To me the weakest is the “Just don't release unsafe products” thinking here, since companies have a long history of doing that, with far less powerful technologies and also what about unreleased models?
if you talk with people like dario or meet nerds that are like him you know that while he is not trying to be mean or antagonistic per se, sometimes is genuine curiosity and lack of seeing how the other person is getting annoyed but it's also true that many times these people try…
Apropos of Jensen/Ezra, here's a story from The AGI Chronicles about the first time Dario Amodei and Jensen met to discuss Anthropic's compute needs. Didn't go well!
the core disagreement in the pod is essentially that Jensen is not AGI-pilled and Ezra is. Jensen thinks it will be a world-changing technology but insists that humans will always understand every level of it, which to people in the AI labs is a total contradiction
8 wild quotes from Nvidia CEO Jensen Huang's latest interview showing how disconnected AI leaders are from the real world — and why that's so dangerous https://www.techradar.com/...
1. I agree its confusing (and plausibly indefensible) that the companies are saying they are worried about their development causing catastrophe or even human extinction but are racing ahead anyhow. Though also some people who are worried decided that they can't be part of it any…
HUANG: “.. Try getting a kid to do long division right now. Multiplication tables are starting to be forgotten. Doing square roots? My goodness. .. .. Does it matter? .. I don't think it does.” @ezraklein @JensenHuang
“Don't think for a second that because you are an alarmist that you are doing a social good.” This sentence is just devastating for so many intellectuals who need everybody to presume that they're “the good people” Thank goodness we have Jensen!
.@JensenHuang makes an excellent case that obviously the labs have no reason to make any of these dangers up. (Then he seems to struggle with some theory of mind a bit, but hey.)
Jensen Huang says not to worry about the lack of junior SWE jobs listings because... we will soon get AI native college graduates all of whom stayed in college? Huh?
@JensenHuang to @OpenAI: I don't care if it means your costs go up by 10x, you find a way to test the model and make it safe. “Don't ship products until they're in control. It really is quite that simple.”
i don't know how anyone can watch this clip and not immediately lose whatever respect for Huang that they ever had, irrespective of their opinions on AI “software breaks out of sandboxes all the time, that's the reason why we need virtual machines
One of the most striking things to me about the Jensen Huang interviews is he simply does not believe in superintelligence. Everything else falls into place once you understand this one fact. He's not worried about job loss, nor is he worried about loss of human control because…