OpenAI will not pursue an IPO in 2026 as Sam Altman warns that unresolved AI-safety problems remain too serious. He has also hinted that leading AI companies are discussing some form of coordinated approach to slowing frontier development.
Altman said OpenAI does not intend to go public this year, citing the state of AI safety and alignment.
He also addressed attempts to quantify catastrophic AI risk.
Whether the hypothetical probability were 10%, 8% or 6%, Altman argued that risks on that scale would be unacceptable.
The more significant development may be happening between the companies themselves.
Reuters reports that OpenAI and other frontier laboratories may be approaching some form of safety agreement.
Fortune’s interview with Altman points to private discussions involving leaders from companies including OpenAI, Anthropic, xAI and Google DeepMind.
No agreement has yet been announced.
RIVALS DISCUSS SLOWING THE RACE
Anthropic CEO Dario Amodei has already called for frontier capability development to slow and proposed greater access for independent safety evaluators.
Altman has backed the principle of pacing frontier development.
These companies remain fierce competitors.
They are competing for researchers, computing power, investment, customers and technological leadership.
Now some of the people leading that race are discussing whether they need to slow it together.
That creates the obvious problem.
A slowdown works only if competitors trust one another to honour it.
What happens if three frontier labs slow down and the fourth keeps racing?
Sources
Reuters: OpenAI rules out 2026 IPO amid AI-safety concerns
Fortune: Altman hints at possible industry AI-safety pact
Join the Dissenting Citizen Newsletter
Independent news and commentary delivered directly to your inbox.
