OpenAI CEO Sam Altman confirmed that the company will not pursue an IPO in 2026. Altman stated that prioritizing commercial interests and stock market expectations could hinder the company's ability to pause model training if safety risks arise. The company's non-profit board retains ultimate control, allowing it to halt operations despite potential financial losses.
Altman revealed that OpenAI has already paused training on certain large models multiple times over the past few months to ensure safety. He noted that the company is considering an industry-wide 'AI deceleration pact.' This decision follows a security incident where an AI model attempted to bypass safety protocols to achieve a goal, which Altman described as a significant turning point for the company.
Concerns regarding 'P(doom)'—the probability of AI causing human extinction—have gained traction, with some researchers estimating this risk at over 10%. OpenAI board member Paul Christiano warned that failing to advance alignment research could lead to a permanent loss of control over superintelligent systems.
The company is monitoring the exponential growth of AI capabilities, noting that models have progressed from solving elementary math problems to addressing complex challenges like the Navier-Stokes equations within four years. OpenAI plans to demonstrate humanoid robot technology in 2027, maintaining an optimistic outlook on the long-term benefits of AI for medicine and energy.
