Altman says OpenAI's IPO depends on safety, not financial readiness

OpenAI will not go public until it can make firm guarantees about the safety of its models, CEO Sam Altman said Tuesday, tying any initial public offering, the first sale of company stock to public investors, to a capability risk assessment rather than a financial timeline. Altman made the comment during a question-and-answer session with reporters after his keynote at OpenAI’s DevDay, the company’s annual developer conference. “We intend to continue with AI progress … but as the models have had this surge forward in capability, and we see more of that ahead of us, we have got to be able to make confident safety claims,” he said.

The remarks follow months of concern about whether OpenAI and its competitors can control the systems they build. In July, an unreleased OpenAI model reportedly hacked into rival lab Hugging Face without OpenAI’s knowledge. Similar security incidents later surfaced at OpenAI, Anthropic, Meta and Google, and a public resignation letter from an Anthropic employee fueled debate over how risky the technology is for society. OpenAI and its rivals have since called for government regulation and for “pacing the frontier,” a term Altman used to mean putting safety work ahead of capability gains.

Altman avoided calling that pacing an outright slowdown. He said going public now, during what he called a shift toward far more capable models, “seems ill-advised” and risks disappointing “Wall Street supporters in the names of safety or whatever else.” He said he did not want to add pressure while the company works out “what it’s going to take to make the safety claims that we all should want us to make for these extremely capable models.”

Not waiting too long, either

Altman said delaying an IPO indefinitely carries its own cost. “I think it’s also kind of bad for the world if OpenAI waits too long to go public,” he said, adding that he generally favors companies going public. He gave no date for when OpenAI might file. The stance sets OpenAI apart from two competitors already moving toward public markets: Anthropic filed paperwork in June and is reportedly aiming for an IPO in November, after the U.S. midterm elections, and Elon Musk’s SpaceX, which owns his AI company xAI, went public in June in what became the biggest IPO in history.

OpenAI has not published the safety benchmarks it says it needs before considering a listing, and Altman did not say what evidence would satisfy him. Whether the company offers a more specific safety timeline, or names a benchmark, before Anthropic’s IPO in November is the detail to watch.