Amid growing concerns about the security risks of artificial intelligence (AI), OpenAI’s CEO Sam Altman confirmed in an interview on Saturday, September 12 that the company will not go public with its initial public offering (IPO) in 2026.
Altman openly stated that given the current security situation, even a 10% risk of AI leading to human extinction before the end of this century is “unacceptable,” and going public now would be “imprudent.”
This stance reflects the deep-seated concerns in the industry about the risks of technological runaway. Recently, two researchers from OpenAI’s competitor, Anthropic, issued strong warnings that rapidly advancing AI technology could trigger human extinction risks in the near future. Anthropic’s CEO Dario Amodei also published an article on Saturday urging AI companies to adopt a more cautious approach in developing technology and actively control the speed of model improvement.
Regarding the discussions about the probability of AI causing human extinction, Altman, in an interview with Fortune magazine on Saturday, stated, “Whether it’s 10%, 8%, or 6%, the key is that all of us have a great responsibility not to let our pride, profit motives, or anything else obstruct us.”
OpenAI’s CFO Sarah Friar had previously revealed that if the business continues to accelerate, the company could go public in 2027 or even earlier. However, Altman made it clear in the Fortune interview that the possibility of going public in 2026 is ruled out, pushing back this highly anticipated IPO to at least 2027.
He emphasized that the industry and the government must collaborate to ensure technical “alignment” and safety. “Alignment” means ensuring that the goals, behaviors, and values of AI systems align with human intent and ethical values during their development, using technical means to prevent AI from exhibiting behaviors that go against human interests or becoming uncontrollable.
OpenAI’s Chief Scientist Jakub Pachocki recently warned that no AI company currently has fully solved the alignment and monitoring issues, and voluntarily slowing down development should be the norm in the industry.
Altman revealed that OpenAI is about to reach a safety agreement with competitors like Anthropic, committing to introduce third-party independent assessors with employee-level authority to implement safety verification.
(This article referenced reports from Reuters and CNBC)