Two companies sit at the frontier of artificial intelligence, David Sacks said, and by his account what they have in the lab is about two generations ahead of everyone else. His answer to their call for a coordinated slowdown is that they can simply do it, today, without asking anyone.
The labs have framed pacing as something that only works if the government convenes it and shields the participants from antitrust law. Sacks' position is that the request is the problem: product safety is already a legal obligation, so trading it for a waiver is bartering over something a company owes anyway.
"Do not demand antitrust waivers or liability waivers in exchange for that."
Sacks chairs the President's Council of Advisors on Science and Technology, which puts him on the government side of the argument he is having with Dario Amodei and Sam Altman in public.
The full interview is covered here so you can skip it.
Here are the 7 arguments that matter.
👤 Guest: David Sacks, Chair of the President's Council of Advisors on Science and Technology
🎙️ Host: Ed Ludlow, Bloomberg Technology anchor
📰 Published: 14 September 2026 on YouTube (Bloomberg Podcasts)
🔴 YouTube | 🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 10 min
Key Takeaways
Frontier AI is a duopoly, and Sacks says the two firms can pace themselves without permission
He puts Anthropic and OpenAI about two generations ahead of the rest
Product liability already makes an unsafe model expensive, so safety is a business decision rather than an altruistic one
He cited Lina Khan saying the major AI companies are subject to it
Asking for an antitrust waiver in exchange for safety is bartering over an obligation the labs already have
The regulation he would accept is transparency and audits — with evaluators who are not funded by the lab's own investors
He calls the AI-extinction debate a hoax and a political operation ahead of the midterms
His comparison is global warming: real concerns blown past their size
1. A Duopoly at the Frontier
Ludlow opened on the labs' own position — that advancement at the frontier needs to be paced — and asked why that does not require a government to impose it.
Sacks' first move was structural rather than philosophical. Look at the market for frontier AI, he said, and "we do have a duopoly situation."
"By any metric, whether it is market share or revenue growth or model capabilities, there are basically two companies at the frontier, and that is Anthropic and OpenAI."
Because the same two firms are asking for the pause, they are also the only ones who need to act. "My point is, guys, go ahead."
He accepts the substance of what they want to do. If the right priority in their product roadmap is to make products more reliable, more trustworthy and more predictable, then they should do it.
"And by the way, they have every incentive to do this."
2. Liability Is the Regulation
The incentive Sacks named is not reputational. It is the exposure a company already carries when it ships something that hurts someone.
He reached for an unlikely ally. "Lina Khan just came out with a statement that I very much agree with, where she said that the major AI companies are subject to product liability."
"So when or if they release an unsafe product, they're massively liable for that." That, he said, makes safety very much in their business interest.
He was explicit that he does not read the labs' move as high-mindedness. "I don't really believe that they're doing this or need to do this for altruistic reasons."
His framing of the duty is deliberately unremarkable. "I think every company in America needs its products when they release them to be safe. That's just a basic corporate responsibility." A company that does not meet it faces civil and, he said, maybe even criminal liability.
He then listed the law that already applies, against the claim that AI is unregulated. "For example, you cannot use AI to conduct a cyber attack. You cannot commit fraud. You cannot commit privacy violations. There's child safety regulations."
"And I'd say most importantly, we have product liability rules."
The conclusion he drew is that the ball is not in Washington's court. "It's on you guys to make sure your products are safe."
3. No Waivers for Doing the Job
The part of Amodei's proposal Sacks objected to hardest is the condition attached to it.
On pacing itself he gave a blessing and a shove. If they feel they need to pace the frontier, "go for it. Do that. You don't need anyone else's permission."
The waiver is where he stopped. "I think Dario basically said in his blog post that he needs an antitrust waiver. Well, no, you don't. You really don't."
He framed a safe-harbor from competition as a benefit no other industry gets. "We don't believe that you should be free from competition." Every company in America would love that deal, he said.
"Do not demand antitrust waivers or liability waivers in exchange for that."
"Honestly, it feels like they're bartering with us, and I don't think they should do that."
His closing threat was reputational rather than regulatory: if the labs cannot make their own products safe, the public will ask why they are doing this at all. "And the public will suggest that they should step aside and make room for people who will do it safely."
4. Two Generations Ahead
Ludlow put the regulatory-capture objection to him: pacing only works if every frontier lab does it together, so should the government get those companies into a room, or keep them out of one? He noted Sacks had raised the cartel question in his own post.
Sacks answered by narrowing the room. On market share, revenue growth and model capabilities, "there's really two companies that dominate it."
"Anthropic and OpenAI are probably about two generations, what they have in the lab is about two generations ahead of everyone else."
"So the others are trying to catch up, but the reality is they have a pretty substantial lead."
He also reframed what is being proposed. He does not think the labs will stop developing: "I think this is just a shift in priorities." The shift is toward reliability and predictability instead of raw power.
With that lead and that liability, he rejects the premise that anyone else has to be constrained for the two of them to act. He does not believe the claim that they cannot make their products safe unless the government imposes requirements on companies that are not even at the frontier.
"Again, this feels like an excuse to me. It feels like bartering. It feels like they are trying to demand concessions from the government in order to meet their basic corporate responsibility. And I just don't buy it."
5. The Panic Is a Political Op
Ludlow asked how far the President wants the opposite — for the labs to keep going — and how that squares with what he had said over the weekend and posted that day.
Sacks split the question in two. What the labs do internally to improve their products is their business: "We're not standing in their way." Separately there is the question of how much new regulation is needed.
The constraint on that second question is China. New rules must not be the reason China leaps ahead.
His example of what he means by unnecessary regulation was a specific bill. "For example, Bernie Sanders has a bill right now that would literally pause AI development and create 20-year prison sentences for AI developers."
"Okay, it's not hard to see that if we go down that road, we're going to lose very badly to our global competitors, basically China."
What he says the President is guarding against is artificial slowing — and the timing is the rest of his answer.
"And what's happened is that the political process in Washington has seized upon a technical debate inside of Silicon Valley and has blown it out of proportion to the point where now people are claiming that the world's going to end in a couple of years."
"And we've seen this movie before." His precedent: "We saw it with global warming. And, frankly, this is a hoax."
"They've taken some legitimate concerns and blown it so far out of proportion that we're into hoax territory here. And this is clearly part of a political op right before the midterms."
Ludlow noted that Speaker Johnson had made the point that day that Congress may need to be an overseer of some guardrail.
6. Audits That Are Independent
Asked where the appropriate place to regulate actually is, Sacks named two things and then attached a condition to both.
"Well, I think we should have transparency. I think we should have audits. I think that is a good idea."
The condition is that an independent evaluator has to be independent, and he said the proposal on the table does not look it. "Dario mentioned a group to be his independent auditor or independent evaluator that was funded by his Series A investors and consists of a lot of ex-Anthropic employees. So is that really independent or is that a cozy arrangement?"
His test is objective and neutral, not merely external. With that, he said, transparency and audits are all in the realm of the positive.
7. China Will Not Pause
Ludlow closed on the part of Amodei's essay that says the US should not slow to the point of falling behind China, and asked whether the debate has already reached that level or is still a warning.
Sacks agreed with the premise and said the answer is partly already known. "I mean, their public officials have already come out and said they're not going to pause." He would still have a conversation with China and see what they are thinking.
His risk is that the American argument itself is the opening. If the political debate in the US is about whether to pause, China will take advantage and pull ahead.
The contradiction he identifies in Amodei's own position is that the framework needs something that will not happen. "But at the same time, it's a contradiction in his argument because he seems to be saying that we need some sort of global agreement in order to make his framework work. And I just think that's unrealistic."
His advice to the lab leaders is to work on what they control: focus on the lab, the engineering and the research, and make the products as predictable and reliable as possible.
That is the corporate duty and also the commercial one. "Enterprises don't want a product that does unpredictable things."
He does not question either chief executive's sincerity. "I think he is sincere in wanting his products to be safe. I think Sam is also sincere in wanting his products to be safe."
What he objects to is the sequencing: Amodei should not be demanding a transnational agreement before doing everything already within his power.
Bonus Insights
Sacks expects the chasing labs to follow the leaders' lead without being told to, saying the other companies will also abide by the same sort of norms and conventions.
His overall reading is that the problem is small and the noise is large. "So, look, I think this is a very manageable problem." What is happening, he said, is that people are seizing an issue a couple of months before an election, and the downside is a lot of unnecessary fear.
"And I think the president is right to try and pop this bubble and point out that this is a panic."
On who the frontier actually belongs to, he was blunt that the two companies are describing themselves: when they talk about pacing the frontier, they are the ones setting it.
Sacks' bottom line is that the two companies at the frontier already have both the lead and the legal exposure to slow themselves down whenever they choose, so the only thing a government needs to supply is transparency and genuinely independent audits — and that anything more, in his reading, is a political panic the administration should deflate rather than legislate.
Products, Companies & Tools Mentioned
Anthropic and OpenAI (The two companies Sacks says make up the frontier duopoly on market share, revenue growth and model capabilities, with what is in their labs about two generations ahead of the field)
Books & Resources Mentioned
We Must Pace the Frontier – Dario Amodei (The essay the whole interview responds to, including the antitrust-waiver request and the independent-evaluator proposal Sacks questions)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: