OpenAI CEO Sam Altman and top executives recently engaged in a Reddit AMA, addressing major concerns about AI competition, open-source policies, pricing, and future developments. The discussion came amid growing pressure from Chinese AI firms like DeepSeek, regulatory scrutiny in the U.S., and OpenAI’s ambitious data center project.

 

Altman acknowledged that DeepSeek’s rapid advancements have narrowed OpenAI’s lead. He admitted that OpenAI’s closed-source strategy might have been a misstep, hinting at a potential shift towards more transparency. However, he noted that not everyone at OpenAI agrees, and it isn’t a top priority right now. Chief Product Officer Kevin Weil suggested the company might open-source older models in the future but provided no specifics.

 

A hot topic was OpenAI’s latest “reasoning” AI models, like o3-mini, which don’t fully disclose their decision-making process to prevent competitors from accessing training data. Weil hinted that OpenAI might reveal more insights into how these models think but stopped short of promising full transparency.

 

On ChatGPT pricing, Altman reassured users that there are no plans for price hikes, and if possible, he hopes to make it more affordable over time. This comes after his previous admission that OpenAI loses money on the $200-per-month ChatGPT Pro plan.

 

Regarding OpenAI’s major data center project, Weil confirmed that more computing power leads to better AI, justifying their massive investment in “Stargate,” a facility expected to require billions in funding.

 

Altman also touched on AI’s potential for self-improvement, acknowledging that a “fast takeoff” scenario—where AI rapidly enhances itself—now seems more likely than he previously thought. He didn’t elaborate further.

 

Concerns were also raised about OpenAI’s partnership with the U.S. government for nuclear defense research. Weil dismissed fears about AI contributing to nuclear weapons development, expressing confidence in the scientists involved.