'If You Can't Control It, Don't Do It': Trump's Top AI Adviser Tells Anthropic and OpenAI to Police Themselves — or Step Aside

The loudest AI safety debate in years just got its sharpest rebuttal — from inside the White House’s own orbit.

In an exclusive interview with CBS News, David Sacks, a key adviser to President Trump on technology and co-chair of the President’s Council of Advisors on Science and Technology, said fears that artificial intelligence could pose an existential threat to humanity are overblown, and that the industry’s recent wave of self-generated alarm is “becoming a panic.”

“We’re in an election season, and there is an incentive for political actors to try and amplify the fear,” Sacks told CBS News senior business and technology correspondent Jo Ling Kent. “And I think this is a moment to really calm down a little bit and look at the issue rationally.”

Safety is the labs’ job, Sacks says

Sacks’ most pointed comments were aimed directly at the chief executives of Anthropic and OpenAI — the two labs whose leaders have spent the past week publicly urging the industry to slow down.

“First and foremost, it is on them to make their products safe,” Sacks said of the frontier model developers. “They should tell us how they should do that. If they can’t do that, they should step aside for people who can do that.”

And in a line that cuts against the entire framing of the slowdown campaign, he added: “What they should not be doing is making the public afraid that they’re doing a bad job.”

The subtext is hard to miss. Anthropic CEO Dario Amodei told CBS News’ “Sunday Morning” that “there are real dangers” with AI and that the “industry lied to people about the fact that this technology had risks.” Amodei, along with OpenAI CEO Sam Altman and other tech leaders, has called on the industry — and governments — to apply the brakes. Sacks’ response to that posture was blunt: “Why are you acting like this is something that you can’t control? If you can’t control it, then don’t do it.”

“Now, look, at the end of the day, I think these risks can be managed, and I do not want to stop. I want to proceed safely,” Sacks said.

Where this is coming from

The interview lands at the peak of a week-long escalation in AI risk rhetoric. On Sept. 8, former Anthropic researcher Jacob Coxon warned that Anthropic and OpenAI are “gambling with our lives” by developing self-improving “superintelligence.” Three days later, Amodei published “We Must Pace the Frontier,” a widely circulated essay arguing that the industry must deliberately slow the rate of capability improvement. Altman has separately named the two ways he believes AI development could go “very bad,” and executives across the sector have joined what has become an industry-wide chorus.

The concern centers on superintelligence — a theoretical stage at which AI systems outperform humans at essentially every cognitive task. Some researchers put that milestone one to 10 years away; others consider the framing itself premature. Sacks falls firmly in the skeptic camp, noting that dire predictions made about AI to date have not come true.

Notably, Sacks speaks with an unusual dual authority here. He stepped down as the president’s AI and cryptocurrency czar in May but remains co-chair of the Council of Advisors on Science and Technology, and he is one of the most influential venture capitalists in Silicon Valley as co-founder of Craft Ventures, with investments spanning Airbnb, SpaceX and Facebook.

The China factor

Underlying Sacks’ reluctance to endorse any slowdown is a competitive argument that has become the administration’s de facto AI doctrine: if the United States pauses, China sprints.

“If we just pause all development, they will race ahead. It’s pretty clear they’re not going to stop,” Sacks said, referring to China, which has rapidly closed the gap with American AI players across open-weight models, chips and applications.

It’s the same argument President Trump made on Monday when he dismissed calls for stronger guardrails and claimed the industry is already sufficiently regulated. And it’s the argument Beijing itself made official this weekend, formally rejecting the Western labs’ “pace the frontier” proposal as a Cold War playbook designed to freeze China’s progress while American labs keep running.

The result is a tidy, three-way standoff: lab CEOs asking governments to slow the race, the U.S. government telling the labs to slow themselves if they’re worried, and China declining to slow anything for anyone.

Why it matters

Strip away the rhetoric and Sacks has articulated the administration’s actual position on AI safety: it is a corporate responsibility, not a regulatory one. That has several concrete implications.

First, it effectively forecloses new frontier-model legislation in the near term. With Congress already stalled on AI action and the White House framing further regulation as unnecessary — and as a strategic gift to Beijing — the voluntary frameworks established by June’s Executive Order 14409, which asks developers to share new models with federal agencies up to 30 days before wider release, are likely to remain the ceiling of federal oversight for this administration.

Second, it puts the labs in an uncomfortable bind. Amodei’s slowdown essay argues that unilateral restraint is insufficient precisely because any lab that pauses unilaterally simply cedes the frontier to less cautious competitors — which is why he asked for government coordination in the first place. Sacks’ answer — that labs which can’t ensure safety should “step aside” — doesn’t resolve that collective-action problem; it restates it as a private choice. A lab that follows Sacks’ logic and slows down loses to OpenAI or to Chinese labs. A lab that keeps racing is, by Amodei’s own admission, outpacing its ability to test and understand its systems.

Third, the political timing is doing real work. Sacks explicitly tied the safety chorus to election-season politics, and the framing will inevitably color how the public parses the next dire warning from a lab CEO. When the people sounding the alarm are also the people selling the technology, and when they ask for the kind of government coordination that would also formalize their incumbency, skepticism is not an unreasonable response — that much, at least, Sacks shares with the industry’s harshest outside critics.

The debate is no longer about whether AI is risky. Even the White House’s own adviser concedes “caution is warranted” and that the risks “can be managed.” The debate is about who pays for the caution — and David Sacks has made the administration’s answer unmistakably clear: not the government, and not by slowing down.