OpenAI CEO Sam Altman has backed a federal framework for consistent safety requirements for frontier AI companies, saying the industry must ensure that safeguards keep pace with the rapid rise in AI capabilities.
Altman said companies should begin developing and applying those safety measures without waiting for legislation or an antitrust exemption. He also said the industry should not treat pacing as a halt to AI development.
"The world deserves confidence that American companies developing increasingly capable AI will act responsibly, especially as the trajectory of progress has steepened," Altman said in a post on X.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
"Every frontier lab must deliver on this, and there is no reason any of us should come to work if we cannot."
Altman said OpenAI welcomes a federal framework that sets consistent safety requirements for frontier AI. But he said companies should start the work of providing that confidence themselves rather than wait for an "anti-trust exemption or legislation".
He said consistent rules for managing frontier risk could help maximise the benefits of AI and added that OpenAI was "excited by ideas like independent auditors".
ALSO READ: AI Could Create Catastrophic Risks By 2030 | Here's What Experts Fear Most
AI Pacing Does Not Mean Stopping
Altman said AI companies had previously developed frameworks such as Responsible Scaling Policies and Preparedness Frameworks. Those measures were useful at the time but focused mainly on the deployment of completed models, he said.
The rapid increase in AI capabilities now requires new tools that focus on safe development and evaluation, according to Altman.
"At OpenAI we now formulate explicit safety cases in advance of frontier reinforcement learning runs we expect to significantly increase capability," he said.
He added that OpenAI would continue the safety work it has done before model releases.
Altman said other AI companies should learn from those approaches and develop their own standards for misalignment, monitoring and safety.
"We look forward to collaborating with our colleagues across the industry to formulate the best version of these," he said.
Altman also sought to clarify what he means by pacing the development of AI.
"When we talk about 'pacing', we do not mean 'stopping'," he said. "Progress has been rapid and will continue to be."
He said AI development should instead proceed at a slower pace when needed to allow safety measures such as safety cases and monitoring to keep up.
"Pacing will be well worth this cost; no amount of American competitive pressure should justify recklessness, or let capabilities get ahead of alignment and monitoring," Altman said.
He said government support would be needed for international coordination, but added that companies should first do what they can themselves.
ALSO READ: AI Slowdown Debate: Satya Nadella Backs 'Embedded Evaluators' - Here's What He Wants
Dario Amodei Calls For Frontier AI To Be Paced
Altman's latest comments follow his endorsement of Anthropic CEO Dario Amodei's call to pace frontier AI development.
Amodei has proposed a three-step framework centred on independent evaluators, coordination among AI companies in democratic countries and global coordination. He said pacing should give companies more time to align and safeguard increasingly capable models while allowing third-party evaluators to verify that work.
The first step calls for every frontier AI company to give a team of third-party evaluators ongoing, employee-like access to verify safety practices, report incidents and assess the alignment of both completed models and training processes. Anthropic has committed to this step.
Amodei has also called for frontier AI companies in democratic countries to establish common safety standards and limits on the rate of unchecked AI progress, with government support where required. His third step calls for global coordination.
Altman said earlier that he agreed with Amodei that the frontier needed to be paced and that OpenAI would adopt the proposal for independent evaluators with employee-like access.
His latest post takes that discussion further by arguing for consistent safety requirements and greater coordination between companies and government.
The central distinction, however, remains the same: pacing AI development is intended to keep safety and monitoring alongside capability growth, not to stop technological progress.
ALSO READ: Can AI Kill Humans? Anthropic CEO Backs Former Employee's Apocalypse Fears
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
