Sam Altman: OpenAI has been on the ‘wrong side of history’ concerning open source

OpenAI researchers, engineers, and executives, including CEO Sam Altman, participated in a Reddit AMA to address questions following a day of product releases.

The company is navigating challenges amidst allegations of IP theft by Chinese companies like DeepSeek, dwindling lead in AI, and the need to reassess its open-source strategy. OpenAI is also exploring the possibility of open-sourcing older models and revealing more about its reasoning models.

Altman and Weil addressed concerns about pricing changes for ChatGPT and the importance of compute power in developing more advanced models. They also discussed the potential for recursive self-improvement and the use of AI models in nuclear defense research.

Technical questions were raised about upcoming models like o3 and GPT-5, as well as updates to DALL-E 3. OpenAI is working on these developments to enhance its image-generating capabilities.