Sam Altman says OpenAI won't go public in 2026 amid AI safety concerns
ChatGPT-maker confidentially submitted an S-1 filing with the SEC in June to begin the IPO process
Sam Altman addresses AI safety concerns with new GPT-6 Astra model
Liz Claman presses OpenAI CEO Sam Altman on the safety of the new GPT-6 Astra model. Altman addresses concerns regarding a previous AI test model breach of Hugging Face and details Astra's strict security standards.
OpenAI CEO Sam Altman signaled that the company will delay its initial public offering (IPO) amid concerns about the safety and alignment of AI models given potential risks they could pose to society.
The ChatGPT-maker confidentially submitted an S-1 filing with the Securities and Exchange Commission in June to begin the IPO process. The company said at the time that it hadn't decided on the timing of the IPO, saying "it may be a while" because it could be easier to do certain things as a private company in the meantime.
OpenAI CEO Sam Altman said concerns about artificial intelligence are understandable following a recent cybersecurity incident involving one of the company's models. (Anna Moneymaker/Getty Images)
Altman said in an interview with Fortune that "we're not rushing into an IPO" and that "I think given everything that's happening with safety… right now would be an ill-advised moment to go public."
MICROSOFT CEO SAYS SUPERINTELLIGENCE MUST REMAIN 'UNDER HUMAN CONTROL'
"We don't feel pressure on that. We've said for a long time, we'll do it when we're ready, which is when the business is ready. When we feel ready from a kind of what the moment is like in society with this technology," Altman said, adding that the IPO won't occur in 2026.
Sam Altman addressed questions on X about OpenAI’s classified work with the Department of War. (Justin Sullivan/Getty Images)
"We've got a lot of stuff to do," Altman told Fortune. "Meeting this moment of what is going to be required for safety and alignment, and how the industry and governments can work together – I'm happy to be able to do that as a private company."
Altman's comments come amid a growing focus by AI companies on ensuring the safety and alignment of their models as they develop, with researchers warning there is a chance an AI superintelligence could wipe out humanity within a decade.
NVIDIA CEO JENSEN HUANG DECLARES ‘AGI HAS ARRIVED’ AFTER OPENAI UNVEILS GPT-6 ASTRA
Last week, former Anthropic and OpenAI researcher Jacob Coxon wrote a lengthy resignation thread posted on X which warned that "the people building AI earnestly believe that it could kill us all by the end of the decade."
Coxon said in his three years doing pretraining research at the two AI giants showed him that "Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.
Anthropic CEO Dario Amodei (Anna Moneymaker/Getty Images)
Evan Hubinger, the lead of Anthropic's alignment science team, responded to Coxon's post saying that "Jacob is correct here – we really do earnestly believe AI could kill all humans!"
"I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and aren't clearly on track to," he continued, adding that while current models aren't a threat, improvements in technology could change that.
"To be clear, as we say in our latest Risk Report, I think the risk from present models is low. What I am worried about is superintelligence arising from recursive self-improvement, as we have said is happening faster than we thought," Hubinger added.
OPENAI SIGNALS POTENTIAL STOCK MARKET DEBUT WHILE WEIGHING PRIVATE-COMPANY ADVANTAGES
Altman said in a post on X that he agrees with Anthropic CEO Dario Amodei that "we need to pace the frontier" of AI development, adding that it "has been a primary topic of discussions we've had at OpenAI in recent weeks."
The OpenAI CEO also signaled he agrees with Amodei's plan to give third-party evaluators employee-level access to Anthropic's systems to verify adherence to safety measures, report on incidents and assess the alignment of models during training – saying that he plans to institute a similar system at OpenAI.
"Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon," Altman said.