Sam Altman warns of two major AI risks

OpenAI CEO Sam Altman has described two ways artificial intelligence could cause serious harm: people could lose control to increasingly capable AI systems, or power over the technology could become concentrated in one person, company or country. He discussed both risks on the Figuring Out With Raj Shamani podcast.

In the first scenario, people could gradually hand decisions to systems they cannot later redirect or stop. Altman said that control could be ceded intentionally or accidentally. The concern is not that AI must act maliciously: systems pursuing an assigned goal could produce outcomes people did not intend, particularly if they are used in critical systems and operate with little human supervision.

The second risk is excessive concentration of AI power. Altman said he favors multiple efforts with different approaches competing, and is wary of too much power in too few hands. He also claimed OpenAI has the most users and the strongest stack, from chips and data centers to models, while saying the company wants others to succeed.

Altman remains optimistic about AI’s potential benefits. He said he trusts people to adapt and believes AI could help researchers discover new science, including insights in physics, medical treatments and energy sources. His comments frame the risks as challenges to manage alongside continued development.