Dossier mode
Read Between the Lines
The same investigation, restaged one beat at a time. Drive it with the arrow keys, space, or autoplay. Nothing is cut from the piece — long runs are split across frames. Read the full investigation or open the The Tech Right hub.
Read between the lines.
Nvidia's Jensen Huang says the AI labs aren't asking to be regulated — they're asking to be relieved of the laws we already have. We take the admission seriously and grade it.
COI up front: this was drafted with Anthropic's model, and Anthropic's CEO is one of the executives named asking for the 'safety' regime we question — so we grade our maker HARDER. The labs DO ask for rules; the real ask is a SWAP — trade the general laws that already make them liable for a bespoke AI regime they help write, sealed with federal preemption.
Huang's quote = FACT. Labs-ask-for-rules = FACT. Existing law already applies = FACT (FTC). The 99-1 preemption fight = FACT. The swap-is-capture read = PROBABLY TRUE, motive POSED not stamped. We do NOT claim all safety concern is insincere or a single conspiracy.
Huang said the AI labs aren't asking for more laws — they're asking to be relieved of the ones we have.
In a CBS News interview (Jo Ling Kent, aired Sept 20, 2026), Nvidia CEO Jensen Huang called the frontier labs disingenuous: 'Go and read between the lines. They're actually not asking for more laws. They're asking to be relieved of the laws we do have.' He added they 'must be doing it for ulterior reasons,' and separately: 'We don't need more regulations. We need to apply the current regulations we have.' Corroborated across Business Insider, RealClearPolitics, and CBS.
The other half: the labs really do ask for rules — so 'they want no oversight' is the wrong charge.
This is the sharpening that keeps the piece from wobbling. Sam Altman told a Senate subcommittee in May 2023 that the government should license the most powerful AI models — he asked to be regulated, on camera. In 2026 Anthropic's Dario Amodei published an essay calling for government action on frontier labs, boosted by Altman and Musk. Any claim that these companies want a lawless free-for-all is false. The real question is which rules — and in exchange for what.
What's on the table is a swap: replace the general laws that already apply with a bespoke AI regime the industry helps write.
The laws that already govern AI are the boring, powerful, general ones: product liability and tort (you built it, it hurt someone, you pay), the CFAA, contract and consumer-protection law, civil-rights statutes, the FTC Act. Regulators say they already apply — FTC Chair Lina Khan: 'there is no AI exemption from the laws on the books,' and an April 2023 FTC/DOJ/CFPB/EEOC joint statement affirmed it. A new AI-specific 'safety' regime — audits, licensing, disclosure standards written with industry input — can quietly swap open-ended liability for a compliance checklist: pass the audit, file the paperwork, you've met your duty. Threat converted into moat. Probably true; some safety concern is sincere; our maker is an advocate (§0).
The tell is federal preemption: a ten-year ban on state AI laws — beaten 99-1, and not over.
If the goal were only more safety, you wouldn't also need to stop states from adding their own. In 2025 a ten-year federal moratorium on state AI regulation was written into the budget-reconciliation bill — a blanket bar on state AI laws. It was so plainly a giveaway the Senate stripped it 99-1 on July 1, 2025, over bipartisan opposition from governors and AGs. The preemption drive continued into 2026. A safety regime you're not allowed to exceed isn't a floor — it's a ceiling, and a ceiling is what an incumbent wants.
Who's saying it matters: Huang is talking against his own commercial interest.
The usual way to dismiss a claim is to find the speaker's angle. Huang's runs the other way. Nvidia sells the chips powering the build-out, and its biggest customers include the very labs — OpenAI, Anthropic — whose executives he just called disingenuous. He's the last person with a commercial reason to embarrass them; keeping the labs spending is Nvidia's whole business. An admission against interest raises the cost of waving him off as a self-interested crank. It doesn't prove the 'ulterior reasons' — we don't stamp that — but it's why the quote is worth building on.
The pattern: 'responsible AI safety' as a cover for trading ordinary liability for rules the incumbents help write.
The same machine this project keeps documenting: a public-interest frame — 'safety' — over a private-interest ask. Not escaping oversight; swapping the open-ended, expensive kind that already binds them for a purpose-built regime that turns legal exposure into a plannable compliance cost, and preempting states so no one raises the bar. Regulatory capture wearing a safety halo. Probably true, not certain — can't read minds, some concern is real — but the shape is consistent: the ask cuts incumbents' liability and raises rivals' barriers while claiming to protect the public, our own maker included.
'We'll write the rules' is not the same as 'regulate us.'
When a powerful industry volunteers to be regulated, the reflex is relief. But there's a world of difference between accepting the laws that already bind you and offering to help draft the ones that replace them. Today's laws are open-ended: hurt someone and a court can make you pay; a state can pass something stricter tomorrow. A bespoke 'AI safety' regime, written with the industry in the room and sealed with preemption, turns that into a checklist — and freezes the rules where the biggest players can afford them and the smaller ones can't. The tell isn't that they asked for rules. It's that they asked to be relieved of the ones that already make them pay, and to stop any state from asking for more. The model writing this belongs to a company on the asking side of that trade.