“Go ahead”
David Sacks, who chairs the President’s Council of Advisors on Science and Technology, said OpenAI and Anthropic are free to slow their own development, but should not make new regulation the price of doing so. “Dario has written that we need to ‘pace the frontier,’ and Sam has agreed. People may be surprised by my response: go ahead,” Sacks wrote on X, ANI reported.
He was responding to Anthropic chief executive Dario Amodei’s essay calling on the leading labs to pace the frontier, which OpenAI’s Sam Altman said he supported. The essay proposes embedded independent evaluators, coordination among democracies and, eventually, global coordination.
Sacks argued the companies had reasons of their own to slow down. “Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack,” he wrote. Trading some raw capability for reliability was “good business”, he said, according to The Tribune.
The objection is to the price
His target was the part of the proposal that needs governments. “The easiest way not to build superintelligence is for you to agree not to build it,” Sacks wrote. Making a preferred regulatory framework the condition for slowing down, he warned, could look like “blackmail”.
The distinction matters because a single lab can slow itself at any time, but the coordination Amodei describes cannot happen without states. Sacks’s answer is that the first part needs no one’s permission, and that the administration does not see the second as its job.
His view lines up with the rest of the administration’s weekend. On Sunday, President Donald Trump said “negative forces” were raising concerns about AI and that he wanted the United States to stay ahead of China, the Associated Press reported.
Meta and Hugging Face take different routes
Alexandr Wang, Meta’s chief AI officer, did not address regulation, but said the company was moving effort into alignment. “MSL is rapidly scaling up the share of our efforts that goes into alignment as our models become more powerful. We do believe alignment can be the gating factor for scaling as we get closer to the frontier,” Wang wrote on X, according to ANI. “People need agents they can trust to reliably do what they ask.”
Hugging Face took Amodei’s proposal at its word. Chief executive Clem Delangue announced an Open Alignment Initiative led by co-founder Thomas Wolf, arguing that alignment “won’t be solved behind the closed doors of a handful of frontier labs”. He asked for Hugging Face to be part of the embedded evaluators programme Amodei had committed to.
What to watch
Two tests follow from Sacks’s post. The first is whether OpenAI or Anthropic announces a concrete limit on its own development without waiting for legislation, which is what he says they are free to do. The second is whether Anthropic accepts Hugging Face into its evaluator programme, which would be the first sign of the proposal opening to organisations outside the frontier labs.
