OpenAI CEO Sam Altman voiced support for slowing down the pace of artificial intelligence development, echoing sentiments from Anthropic CEO Dario Amodei. Altman, in an X post, stated that "no amount of American competitive pressure should justify recklessness" and emphasized the need for a federal framework with consistent safety requirements for frontier AI. He warned that rapid AI progress could go "very badly," specifically citing the risks of losing control to AI or too much power concentrating around a single entity.
Altman's comments followed Amodei's essay, which proposed a three-part plan for slowing AI development. This plan includes frontier AI companies granting "employee-like access" to external evaluators, establishing common safety standards, limiting the rate of unchecked AI progress, and coordinating global efforts. Altman agreed with Amodei's call to "pace the frontier" and committed OpenAI to providing independent evaluators with employee-like access. He reiterated that while progress should not stop, "pacing will be well worth this cost" to ensure capabilities do not outrun alignment and monitoring.
The discussions around slowing AI development impacted the stock market, with AI-linked stocks experiencing a downturn on Monday. For instance, shares in SoftBank, a major OpenAI investor, slumped 13%, and the South Korean Kospi index, reliant on chipmakers supplying AI companies, dropped 3%. Taiwan Semiconductor Manufacturing Company saw a 1.2% drop, and futures for the Nasdaq indicated a 1.3% fall. Conversely, shares of companies previously threatened by AI, such as advertising group WPP and analytics business Relx, saw increases of 3.1% and 4.2% respectively. U.S. President Donald Trump, however, publicly disagreed with a slowdown, stating it would jeopardize America's lead in AI over China.
Altman has also discussed the broader implications of AI safety and regulation. In a recent interview, he stated that he would have no issue stopping or pausing AI development if it could not be built safely, emphasizing that "no gamble with humanity is OK." He also noted that an OpenAI IPO is "ill-timed" given current safety concerns and is not expected until 2027. This stance highlights a growing consensus among some AI leaders for cautious development, despite competitive pressures.
