OpenAI CEO Sam Altman says one of his biggest concerns about AI is not whether it becomes too powerful in and of itself, but whether a single company could control that power.

“I think the struggle of the moment is: Are we heading into a world of authoritarianism or freedom in AI?” Altman said on the latest episode of the “Relentless” podcast.

“Every time humanity has traded its freedom for security, it has been a net long-term loss,” Altman said. “We’re going to put this in the hands of the people. We’re going to empower them. We’re going to let society express their ideas and use this technology the way they want.”

It is a surprising argument from the director of OpenAI, one of the few companies that control the most powerful AI models in the world. OpenAI is also proprietary and not open source, meaning it does not share its technology.

Altman’s comments are timely. They follow a debate that erupted in Silicon Valley and Washington last week over how much visibility the general public should have about the architecture of frontier models.

The recent launch of Moonshot AI’s Kimi K3, a powerful open-source Chinese model, prompted calls from some US officials to restrict access to Chinese models, and warnings from investors and startup founders that doing so could weaken global competition.

The gap extends beyond access to models. US export controls already determine which countries and companies can get the most advanced chips and computing power, giving Washington significant influence over who can build cutting-edge AI.

Microsoft, Meta, Nvidia, OpenAI, Google and many other leading tech companies signed a letter urging Washington not to restrict open source AI.

Anthropic, OpenAI’s biggest competitor and arguably the leading AI company on the planet right now, was the only major frontier lab that did not sign the letter or publicly endorse its contents. The company is known for taking a more cautious approach, arguing that increasingly capable models require stronger safeguards and stricter controls.

Meanwhile, Altman said preventing that power from becoming overly concentrated has long been central to his vision of AI, even as his company continues to maintain its own closed models.

“I have long felt focused on this singular goal of abundant intelligence and the belief that incredible human prosperity will come from it, as long as we don’t have a strange concentration of power and a kind of new authoritarianism,” he said.