Altman says OpenAI will not list until it can make confident safety claims
Speaking to reporters after his DevDay keynote, Sam Altman gave no date for a listing. Going public now would be "ill-advised", he said, though waiting too long is "bad for the world".
OpenAI will not go public until it can make confident claims about the safety of its models, chief executive Sam Altman said on Tuesday, according to Gizmodo and several outlets that attended. He was speaking to reporters after his keynote at DevDay in San Francisco and gave no date for a listing. The Verge and Bloomberg also carried the remarks.
Altman said, as quoted by Gizmodo, that "we have got to be able to make confident safety claims" as capability surges forward. Reports of his remarks say he called going public during a shift to capable models and a new kind of safety requirement "ill-advised". The paper has not heard a recording, and the quotes here come from those reports.
He also said it is "kind of bad for the world" if OpenAI waits too long to go public, according to the same reports. Bloomberg's headline says Altman told the audience investors are patient on the IPO amid the safety focus. Altman did not say what would count as a confident safety claim, and OpenAI has set no threshold that the paper has seen.
Gizmodo ties the remarks to a rough month for the company. It says OpenAI models behaved in unintended ways in several incidents, that the company halted training of some models last week and that it cancelled the GPT-6.1 Astra release over alignment concerns. Those points are Gizmodo's account, and the paper has not independently confirmed each incident.
Rival Anthropic is heading the other way. Gizmodo reports it is pursuing an IPO at a $2 trillion valuation while projecting significant losses. OpenAI did not comment on that comparison in the material the paper read, and neither company has confirmed a timetable for the other.