“We joined OpenAI because we wanted to ensure the security of the incredibly powerful AI systems the company develops,” researchers William Saunders and Daniel Kokotajlo wrote in the letter. “But we left OpenAI because we lost confidence that the company would develop its AI systems safely, fairly, and responsibly.”

Saunders and Kokotajlo did not immediately respond to requests for comment from Business Insider.

They continued: “Developing cutting-edge AI models without adequate safeguards poses foreseeable risks of catastrophic harm to populations.”

OpenAI CEO Sam Altman has repeatedly and publicly expressed his support for the concept of AI regulation, Saunders and Kokotajlo wrote, citing Altman’s congressional testimony in which he called for government intervention, but “when regulation actually occurs, he opposes it.”

An OpenAI spokesperson told BI in a statement: “We strongly disagree with the misrepresentation of our position on SB 1047.” The spokesperson referred BI to a separate letter that OpenAI Chief Strategy Officer Jason Kwon wrote to California Sen. Scott Wiener, who introduced the bill, detailing the company’s concerns.

SB1047 “has inspired thoughtful debate,” and OpenAI supports some of its security provisions, according to Kwon’s letter, dated a day before the researchers’ letter was sent. However, because of the national security implications of AI development, the company believes regulations “should be shaped and implemented at the federal level.”

“A government-driven AI policy, rather than a patchwork of state laws, will spur innovation and put the U.S. in a leading position in developing global standards,” Kwon’s letter said.

However, Saunders and Kokotajlo aren’t convinced that the push for federal legislation is the only reason OpenAI is opposing SB 1047 in California. They argue that the company’s complaints about the bill are “not constructive and not made in good faith.”

“We can’t wait for Congress to act — they’ve explicitly said they’re unwilling to pass meaningful AI regulation,” Saunders and Kokotajlo wrote. “If they ever do, it could preempt CA legislation.”

The former OpenAI employees concluded, “We hope that the California Legislature and Governor Newsom will do the right thing and sign SB 1047 into law. With the right regulations, we hope that OpenAI can still fulfill its mission statement of building AGI safely.”

Representatives for Wiener and Newsom did not immediately respond to requests for comment from BI.