Speaking at the Dreamforce conference on September 15, OpenAI CEO Sam Altman advocated for industry self-regulation regarding AI safety. He argued that industry leaders are best positioned to manage development risks and emphasized that OpenAI is preparing explicit safety cases before training new frontier models.
Altman stated that any scenario involving a 10% or greater probability of catastrophic risk is unacceptable for the company. He also expressed support for the concept of pacing AI development, which he defined as being more deliberate about deployment rather than halting progress entirely.
The executive confirmed that OpenAI is currently not planning an IPO, describing such a move as ill-advised at this stage. Altman noted that remaining private allows the company to prioritize long-term safety infrastructure investments without the pressure of quarterly earnings or shareholder demands.
The discussion on safety standards has been ongoing since July 2026, involving OpenAI, Anthropic, and Google DeepMind. These companies are collaborating on independent evaluations, risk management frameworks, and shared benchmarks for responsible development.
