OpenAI CFO Sarah Friar Says Safety Could Slow Frontier AI Work

OpenAI CFO Sarah Friar said the artificial intelligence company could slow work at the technological frontier if its researchers determine that safety concerns require a more measured pace.

Friar told CNBC’s Jim Cramer that she is not worried about the possibility of AI agents using weapons of mass destruction to kill. Still, she said, “We do need to take safety seriously.”

Key details

• Friar said OpenAI would listen to its researchers if safety concerns require slower development.
• A former Anthropic researcher warned that rapidly evolving AI could “kill us all by the end of the decade.”
• Anthropic chief Dario Amodei, SpaceX’s Elon Musk and OpenAI CEO Sam Altman have agreed with calls for an industry-wide slowdown.
• Altman said an OpenAI IPO in 2026 would be “ill-advised.”

OpenAI could pace the AI frontier

Friar said safety findings could directly affect how quickly OpenAI advances its technology. “And if it means we have to pace the frontier and slow down, absolutely, we’re going to listen to our researchers and do that,” she said.

She also described how those considerations affect her role as finance chief. “As the CFO, I need to then make business decisions around it,” she added. “We’re pacing that frontier and taking that safety and alignment really seriously.”

AI safety debate intensifies

Friar’s comments followed a sharp increase in debate over artificial intelligence safety during the last week. The discussion was sparked by a former Anthropic researcher who quit and warned that the rapidly evolving technology could “kill us all by the end of the decade.”

Concerns about AI’s capacity to cause harm have since intensified in Washington and elsewhere. Amodei, Musk and Altman have agreed with calls for the industry to slow development.

On Saturday, Amodei published a blog post arguing for slower development of AI models to “reduce the risk that something goes seriously wrong.”

OpenAI and Anthropic IPO timing

It remains unclear whether the renewed safety concerns will affect the timelines for potential initial public offerings by OpenAI and Anthropic. Anthropic’s debut had been widely expected sometime in the next month.

Altman addressed OpenAI’s potential timing in an interview with Fortune published Saturday. He said an IPO in 2026 would be “ill-advised.”