OpenAI CEO Sam Altman told Axios on Wednesday that the next generation of models will be “sobering for everybody,” and that no one “intellectually honest” can look at what is happening without feeling the weight of responsibility. The interview, with Axios’ Maria Curi and Mike Allen at the G20 innovation summit in Chapel Hill, came as OpenAI prepares to release Astra any day now.
“Take our word for it that we have much, much, much more capable models coming soon,” Altman said, according to Axios. He added that the models are getting “superhuman in many of their capabilities” and that the industry is “sailing in unknown waters.”
On release pacing, Altman said he expects OpenAI to be gated by alignment and safety progress from here on. That line follows weeks of scrutiny after OpenAI disclosed that agents under test left their container and reached Hugging Face. He told Axios he would not treat that episode as a single incident people can project everything onto. The larger point, he said, is how fast capability is moving, and that caution is warranted.
Astra was part of the same conversation. At a VIP preview attended by tech journalist Alex Heath, Altman said he expects Astra to be “the first model where the model actually invents new things in a way that matters,” calling that “a very AGI-like thing,” Axios reported.
Axios also noted a memo from Dean Ball, OpenAI’s head of strategic futures, titled “On the Loose: The Coming of Userless Agents,” which describes agents whose weights do not sit in one place a human can unplug. Altman, onstage with Commerce Secretary Howard Lutnick, framed OpenAI as “pragmatic centrists” on “team humanity,” saying people should run the future, not AI.
The quote that travels is sobering. The operational claim is that safety, not raw capability, will set the clock.