What changed: Altman formally delayed the previously reported IPO target, citing AI safety as the primary justification.
OpenAI CEO Sam Altman made critical statements about AI safety and the company's future in an interview conducted by Fortune editor in chief Alyson Shontell. The conversation took place amid the fallout from a cyberattack involving OpenAI and HuggingFace, as broader discussions about AI safety were intensifying.
Responding to a direct question about IPO plans, Altman said this would be an 'ill-advised moment' to go public. He confirmed that the company had filed confidentially for an IPO but made clear that OpenAI would not list on public markets this year. Altman tied the timing of any potential public offering to the company's readiness and to society's relationship with the technology, speaking against the backdrop of earlier reports suggesting a delay due to market volatility and financial challenges.
On the question of safety, Altman openly acknowledged that it was 'absolutely possible' to build an AI that goes beyond human control. He pledged to take all necessary steps to prevent such an outcome, including pausing training. Altman also emphasized that there are risks OpenAI should not be able to incur on behalf of humanity.
Key facts
- Altman confirmed OpenAI will not go public this year, citing safety concerns.
- The company has already completed a confidential IPO filing.
- Altman openly acknowledged that it is possible to build AI beyond human control.
- He pledged to pause training if necessary and noted that some risks should not be incurred on behalf of humanity.
- The interview was given to Fortune amid an agenda dominated by the OpenAI-HuggingFace cyberattack.
Why it matters
A public declaration by the CEO of the world's most influential AI company that he is open to pausing training marks a significant threshold in the industry's safety discourse.
Watch next: OpenAI's IPO timeline and the concrete steps the company takes on AI safety policy will be closely watched.
Source date: 2026-09-14T00:21:04.538Z · Verification confidence: 85/100
