Sam Altman, CEO of OpenAI, said in a podcast on August 23 that he is worried AI will eventually be controlled by a small number of companies, models or individuals, leaving the vast majority of consumers without a voice in how technology shapes the world. He advocated for "the right approach is to let people have deep control over the future, with society and models evolving together."

"Trading freedom for security" could lead to monopolies

Altman pointed out another contradiction: he is also concerned about AI getting out of human control, but his fear of this risk may actually lead to the first scenario. He implied at Anthropic's CEO Amode, who has repeatedly warned that the Mythos model poses a "very real threat" to cybersecurity, finance, and critical infrastructure, and has called for government regulation. In Altman's view, this is "trading a lot of freedom for security."

The disagreement focuses on the "open weights" approach: Anthropic was the only major company that did not sign a joint letter supporting open weights last month, while China has used this to follow Silicon Valley at low cost. NVIDIA, Meta, and Microsoft have also called on the United States to follow suit. Although OpenAI has not open-sourced its cutting-edge models, it still released two open-weight models, gpt-oss-120b and 20b, last year. Altman also said that U.S. AI leaders have not clearly explained to the public how the technology empowers people's autonomy in life, and he believes AI is creating the largest wave of small business startups in history.