Greg Brockman Reveals Inside Story of OpenAI's 72-Hour Near-Collapse When Sam Altman Was Fired
Key Takeaways
- ▸Greg Brockman recounts the 72-hour crisis following Sam Altman's firing, including the emergency 'Phoenix' backup company designed at Altman's house and the moment Ilya Sutskever's tweet changed everything
- ▸OpenAI has operated on a three-step technical plan established at a 2015 Napa offsite and moved away from a pure nonprofit structure due to practical business requirements
- ▸A significant and growing portion of OpenAI's codebase is now written by AI, reflecting the company's advanced capabilities in code generation
Summary
In a rare interview on The Knowledge Project Podcast, OpenAI co-founder and President Greg Brockman provides an unprecedented account of the three-day crisis that nearly destroyed the company when Sam Altman was fired from the board. Brockman details the emergency backup plan, codenamed 'Phoenix,' designed at Altman's house, and how Ilya Sutskever's pivotal tweet reversed the company's trajectory. He also reveals OpenAI's foundational technical blueprint from a 2015 Napa offsite that has guided the company for a decade, and explains the decision to abandon its pure nonprofit structure. Beyond the crisis narrative, Brockman discusses the current state of AI development, including how much of OpenAI's own codebase is now written by AI, why ChatGPT stopped showing reasoning traces, and addresses broader concerns about the global AI race, compute constraints, and AI's impact on employment.
- Brockman addresses OpenAI's decision to stop showing reasoning traces in ChatGPT, the implications of compute-constrained access to AGI, and the impact of AI on employment
