OpenAI has made headlines recently by hiring two cofounders from the Thinking Machines Lab, Barret Zoph and Luke Metz, after they had departed OpenAI in late 2024. This move has sparked speculation surrounding the resignations and the internal dynamics at Thinking Machines Lab.
A source with inside knowledge revealed that the leadership at Thinking Machines considered Zoph’s actions to involve serious misconduct, which allegedly eroded Mira Murati’s trust in him and disrupted their working relationship. It was reported that Zoph was fired by Murati the same day OpenAI announced his rehiring, having raised internal concerns about whether he shared confidential information with competitors. Zoph has not publicly commented on these allegations.
In his memo to OpenAI employees, CEO of applications Fidji Simo indicated that the rehiring of Zoph and Metz had been planned for weeks, asserting that Zoph had informed Murati of his intentions to leave Thinking Machines the Monday prior to his dismissal. He also stated that OpenAI does not share the ethical concerns raised by Thinking Machines regarding Zoph.
In addition to Zoph and Metz, Sam Schoenholz, another former OpenAI researcher who was working at Thinking Machines, is also rejoining OpenAI. Reports suggest that at least two more employees from Thinking Machines are expected to join OpenAI soon.
Amid these changes, some insiders argued that this situation reflects broader misalignments regarding the vision for projects at Thinking Machines, not solely centered on Zoph. The AI sector continues to experience volatility, echoing similar upheavals from previous years, including the brief ousting of OpenAI’s former CEO Sam Altman in 2023.
Concern over the constant turnover and drama in AI labs has led to frustrations among researchers who did not anticipate their industry would face such scrutiny. However, the significant financial backing for AI projects indicates that these types of shake-ups might persist as research and competition intensify.
On a parallel front, AI labs are ramping up efforts to create economically viable agents, revising strategies for training AI technologies to handle complex tasks. Recently, OpenAI has begun collaborating with contractors to gather genuine work examples from various professional sectors, which could enhance the performance and applicability of their AI agents in real-world contexts.
Interestingly, this increment in sophistication may lead to significant developments in how AI is integrated into various industries, including consulting, healthcare, and finance. AI’s progression remains a point of contention, as researchers express weariness over the tumultuous landscape yet recognize the potential advancements that could arise from ongoing investment and innovation in the field.
