OpenAI's personnel turmoil continues, with President Greg Brockman announcing an extended leave of absence. Greg is considered a staunch supporter of Sam Altman and stood by him when Sam was ousted from the board. Although Greg stated he would return to the company after his leave, this behavior inevitably leads to speculation about other possibilities.
Additionally, two core executives have reportedly resigned:
-
Co-founder Johnson Schulman - briefly served as head of the safety team, previously co-led by former Chief Scientist Ilya Sutskever and researcher Jan Leike. Johnson also led the post-training process of the GPT series large models and was a key figure in technical development. Notably, he joined OpenAI's main competitor Anthropic immediately after leaving.
-
Product Head Peter Deng - joined OpenAI just last year and left in less than a year, sparking speculation about the state of OpenAI's team.
After these changes, only Sam Altman, Wojciech Zaremba, and Greg Brockman (who announced an extended leave) remain at the company out of the original 11 co-founders of OpenAI. The rest of the founding members have left the team.
2024 has become the biggest wave of veteran personnel changes for OpenAI since the Musk incident. Departing employees typically choose to join competitor companies (such as Deepmind and Anthropic) or start their own businesses. This has also given rise to the so-called "OpenAI Mafia," with nearly 30 former employees founding their own AI companies, including some unicorns.
The three individuals who resigned this time are all core managers of OpenAI, responsible for the company's core technical work. Although their departures may have limited direct impact on current new model development, they have affected team morale internally and external confidence in the company.
A key term in these resignation events is "superalignment." OpenAI's superalignment project was established in June 2023, aiming to address the alignment of higher-level intelligent systems with human goals. However, this project seems to conflict with Sam Altman's preferred "effective accelerationism" ideology.
Both Jan Leike and Johnson Schulman joined Anthropic after leaving, a company founded by former OpenAI core employee Dario Amodei. Anthropic's Constitutional AI concept shares similarities with the superalignment idea, both emphasizing the use of AI systems to supervise other AI systems to improve AI safety and usefulness.
These resignation events reflect possible disagreements within OpenAI regarding AI development direction and safety issues, and have also sparked external attention and discussion about the company's future development.