OpenAI's CEO, Sam Altman, admits that the company's strategy of keeping its AI technology locked away may have been wrong, as China's DeepSeek gains traction with its open-source AI models. Altman acknowledges that they may have been on the wrong side of history and reveals that his team is discussing a different open-source strategy. OpenAI's chief product officer, Kevin Weil, suggests open-sourcing older models, but there is skepticism surrounding the specifics of this potential change. Altman also acknowledges that OpenAI's models lack transparency, unlike DeepSeek's models, which show the entire thought process. OpenAI is working on increasing transparency while balancing the risk of competitors reverse-engineering their systems. Altman denies rumors of a possible price hike for ChatGPT and expresses a desire to make it cheaper. However, OpenAI is already losing money, and the $200-per-month Pro plan is not profitable. Altman also discusses the company's need for more compute power and addresses concerns about AI's role in weapon development and the concept of recursive self-improvement. Despite these challenges, Altman provides updates on upcoming tech, including the next reasoning model and a successor to its image-generation model.



Other News from Today