Sam Altman: OpenAI has been on the ‘wrong side of history’ concerning open source
Altman discusses OpenAI's stance on open source amid competition.

Sam Altman confessed that OpenAI has fallen behind competitors like DeepSeek and admitted to being on the 'wrong side of history' with their closed-source philosophy. He proposed reevaluating OpenAI's approach to open sourcing older models as a potential avenue for staying competitive. Additionally, Altman mentioned that the race with DeepSeek might prompt OpenAI to disclose more about their AI models' reasoning processes and methods.
Meanwhile, Kevin Weil refuted rumors of a ChatGPT price increase and emphasized that OpenAI is committed to making the platform more affordable. He mentioned initiatives like the Stargate data center project that aims to enable the development of more high-performing models. Weil noted that evidence continues to support the notion that greater computational power leads to significant model improvements.
In a speculative yet cautious tone, Altman discussed the potential for recursive self-improvement in AI models, suggesting he sees 'fast takeoff' as more likely than before. OpenAI's collaboration with the U.S. government for nuclear defense research stirred questions about using AI in developing destructive weapons. Weil expressed trust in the responsible use of OpenAI's models by government scientists.