Bloomberg Tech

David Sacks on why the labs don't need permission to slow down

David Sacks· Chair of the President's Council of Advisors on Science and Technology at The White House
·~10 min·English·Bloomberg
PolicyAI SafetyAI CompanyBusiness Strategy
TL;DR

David Sacks argues the two frontier labs should meet their own existing legal duty to ship safe AI instead of asking the government to bless a coordinated slowdown, and reads the push for new rules as a pre-election panic the US can't afford against China.

01The Core Reply

"Do It Yourselves"

The labs asked Washington to bless a coordinated slowdown; Sacks's answer is that slowing down is their own product decision, not a policy they need granted.

You don't need anyone else's permission.

— David Sacks, Bloomberg Tech
Key Insight
By calling the slowdown a decision rather than a policy, Sacks quietly denies the labs any standing to negotiate. If no permission is required, there is nothing for the government to grant — and nothing to trade for.
02Market Structure

It's Really Just Two Companies

Sacks grounds the whole argument in one fact: Anthropic and OpenAI are the frontier, with a lead he puts at about two model generations over everyone else.

By any metric, whether it is market share or revenue growth or model capabilities, there are basically two companies at the frontier.

— David Sacks, Bloomberg Tech
Key Insight
The duopoly claim does double duty. It says the two labs are big enough to fix safety alone, and it warns that new frontier rules would mostly bind the chasing companies that never set the frontier in the first place.
03The Existing Rulebook

The Rules Already Exist

Sacks lists the laws already covering AI — cyberattacks, fraud, privacy, child safety, and product liability — to argue the incentive to ship safely is built in, not missing.

You cannot use AI to conduct a cyber attack. You can't commit fraud. You cannot commit privacy violations.

— David Sacks, Bloomberg Tech
Key Insight
Leaning on product liability reframes safety from a favor the labs grant into a duty they already carry. Citing Lina Khan — an FTC figure from the other party — is deliberate: it lets a Republican official borrow a bipartisan hammer.
04The Trade He Rejects

Stop Bartering

Sacks says a lab pointing to Dario Amodei's call for an antitrust waiver is trying to trade its basic obligation for a government concession.

Honestly, it feels like they're bartering with us.

— David Sacks, Bloomberg Tech
Key Insight
Framing a waiver request as bartering flips the moral posture. It casts the labs as seeking relief from competition, not seeking safety — so any concession looks like a giveaway rather than a reasonable term.
05Accountability

Or Step Aside

If a lab claims it can't make its own product safe without new rules, Sacks says the public should ask why it is building the product at all.

the public will suggest that they should step aside and make room for people who will do it safely.

— David Sacks, Bloomberg Tech
Key Insight
This turns the labs' own safety worry against them. Admitting you can't ship safely without help stops being a plea for regulation and becomes an argument for why you may not belong at the frontier.
06Where He Would Regulate

Audits, But Actually Independent

Sacks backs transparency and audits — with the catch that an evaluator funded by a lab's own investors and staffed by its ex-employees may not be independent.

Is that really independent or is that a cosy arrangement?

— David Sacks, Bloomberg Tech
Key Insight
It is the one place Sacks concedes ground — and even there the real target is capture. Endorsing audits while questioning the auditor lets him support oversight in principle while resisting the specific institutions the labs would shape.
07The Geopolitics

Don't Hand China the Lead

Sacks warns that a self-imposed pause would let China pull ahead, and calls a proposal for a global agreement unrealistic when China has already said it won't stop.

Bernie Sanders has a bill right now that would literally pause AI development and create 20 year prison sentences for AI developers.

— David Sacks, Bloomberg Tech
Key Insight
The China card recasts a safety debate as a race the US could lose by choice. It also names the contradiction Sacks sees in Amodei's position: a plan that only works with a global pause is, in a world where China won't pause, a plan that can't work.
08The Meta-Read

This Is a Panic

Sacks reads the whole slowdown push as a Silicon Valley technical debate inflated by pre-midterm politics into an end-of-the-world scare.

we saw it with global warming. And, you know, frankly, this is a hoax.

— David Sacks, Bloomberg Tech
Key Insight
Reaching for the global-warming comparison is a tell about audience: it sorts the whole AI-risk conversation into a familiar partisan bin. The move settles the framing without engaging the specific safety claims on their merits.