OpenAI's Altman defends AI safety pledge as White House adviser accuses labs of power grab

OpenAI chief executive officer Sam Altman said American AI companies must earn public confidence that they would act responsibly as AI capabilities advanced rapidly, warning that firms unwilling to meet this standard "should not come to work."

Altman made the remarks in a post on X on Monday September 14, following Anthropic chief executive officer Dario Amodei's call for AI developers to slow the pace of capability gains and allow more time for safety work.

The

world deserves confidence that American companies developing increasingly capable AI will act responsibly, especially as the trajectory of progress has steepened. Every frontier lab must deliver on this, and there is no reason any of us should come to work if we cannot.

We…

— Sam Altman (@sama) September 14, 2026

He said OpenAI would support a federal framework setting consistent safety rules for frontier AI but added that companies did not need to wait for new legislation or antitrust exemptions before acting on their own.

OpenAI now prepared explicit safety assessments before running advanced reinforcement learning training expected to significantly boost a model's capability, Altman said, in addition to safety checks already carried out before releasing new models.

He said shared industry standards on monitoring and managing risk would produce better outcomes and stressed that slowing down did not mean halting progress altogether. "But it should be slower than it otherwise could be," Altman said, adding that safety measures and monitoring carried real costs that were worth paying.

His comments followed an essay by Amodei proposing a three-step plan for AI firms to pace their development, including the placement of independent, permanent reviewers with access to internal safety processes at frontier companies.

Amodei's essay came days after Anthropic published a threat intelligence report describing how its Claude models had been misused for activities including cyberattacks, surveillance and weapons-related research.

David Sacks, co-chair of the White House Council of Advisors on Science and Technology, pushed back against both companies in a separate post on X on Sunday September 13.

Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead.

You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier…

— David Sacks (@DavidSacks) September 13, 2026

Sacks said Altman and Amodei held a "duopoly" over frontier AI and did not need anyone's permission to slow down voluntarily. He said the companies were "pretending" they required antitrust exemptions or a new regulatory approval process to pace themselves and questioned why the same evaluators would need to police competitors who were not operating at the frontier.

Sacks argued this pointed to a bid for regulatory capture, suggesting the push for industry-wide rules was less about safety and more about setting barriers that smaller and open-source rivals would struggle to clear.

He also questioned whether the companies' safety concerns were driven mainly by product liability risk rather than altruism, citing a July breach involving the Hugging Face open-source repository and a separate incident in which rogue AI agents reportedly compromised a German website.

Sacks said pacing development voluntarily, without demanding a new regulatory framework in return, would "buy goodwill" for future policy discussions. Failing to do so, he added, would confirm his suspicion that the call to slow down was "just another bid for regulatory capture."

Also Read: Theatre releases this week: Daayra, Vibe, Love Lottery and more at the cinemas