As AI leaders including Dario Amodei and Sam Altman propose slowing down of advanced AI models amid the growing safety concerns, former White House AI adviser and crypto czar David Sacks has slammed OpenAI and Anthropic over these calls. Sacks argued that the two tech giants are already at the forefront of the AI race and can easily choose to slow down without waiting for government approval.
Sacks was responding to comments from Anthropic CEO Dario Amodei about the need to “pace the frontier”, with OpenAI CEO Sam Altman also backing the idea. Instead of opposing an AI slowdown, Sacks said the companies should simply go ahead if they believe their latest systems are becoming too risky.
‘Stop Pretending You Need Permission’
Sacks argued in a post on X that OpenAI and Anthropic have become the leading players in advanced AI, pointing to their market share, revenue growth and model capabilities. “You guys are the frontier,” Sacks wrote, adding that the two companies have what he described as a “duopoly on frontier intelligence,” tweeted Sacks.
He also referred to claims from the companies that their lead is growing because increasingly capable AI models can help improve the development of future systems. Sacks said that if OpenAI and Anthropic genuinely believe their unreleased models could be dangerous, he supports their decision to slow down.
However, Sacks strongly objected to the idea that the companies need a new regulatory framework to make that decision. “Stop pretending you need anyone else’s permission,” he said.
He also criticised what he sees as attempts to change antitrust rules or create special regulatory arrangements around the development of frontier AI. Sacks argued that OpenAI and Anthropic should not use AI safety concerns to justify policies that could reduce competition.
Moreover, he has also questioned the independence of METR, an organisation involved in evaluating AI capabilities and risks, pointing to its links with people connected to Anthropic.
Notably, Sacks challenged the idea that slowing down AI would necessarily be driven only by safety concerns. According to him, OpenAI and Anthropic have a business reason to make their models more predictable and reliable. He pointed to potential legal and financial risks if an AI system enables a serious cyberattack or behaves in an unauthorised way.
Sacks On AI Regulation And China
Sacks said that slowing the development of frontier AI could also give policymakers more time to discuss how the technology should be regulated. He contrasted this with what he described as Bernie Sanders’ approach of wanting to “shut it all down”.
The former White House adviser also warned that any global agreement on AI would have to consider China, which he said is unlikely to participate in such an agreement. In his concluding argument, Sacks returned to what he sees as the simplest solution: if OpenAI and Anthropic believe the technology needs to slow down, they can make that choice themselves.
“So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it,” he wrote.
He concluded, “If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.”