Meta might have managed to poach researchers with multi-hundred million pay packages, but it’s finding that it’s harder to keep them there.
Former OpenAI researcher Shengjia Zhao, one of the prominent researchers behind ChatGPT, had threatened to return to OpenAI within days of joining Meta, FT reports. Zhao even signed employment paperwork to return to OpenAI. But not long after, he was given the title of “Chief AI Scientist” at Meta, after which he decided to stay with the company.
Some other new hires are also finding it hard to stay at Meta. Ethan Knight, who’d joined the company just weeks ago, has left, while former OpenAI researcher Avi Verma went through Meta’s onboarding process but never showed up for his first day. A day prior to that, Rishabh Agarwal, a research scientist who started at Meta in April, announced his departure. He said that while Zuckerberg and Wang’s pitch was “incredibly compelling”, he “felt the pull to take on a different kind of risk”. Meanwhile, Chaya Nayak and Loredana Crisan, generative AI staffers who had worked at Meta for nine and 10 years respectively, are among the more than half a dozen veteran employees to announce they are leaving in recent days.
All this tumult follows Meta’s failed Llama-4 model, which severely underperformed compared to expectations. As Chinese labs took the lead in the open-source space which Meta had previously championed, CEO Mark Zuckerberg reorganized the entire division, setting up a Superintelligence Lab under Scale CEO Alexandr Wang after acquiring a large stake in his company. Wang, 28, is now leading all of Meta’s AI efforts, and even has veterans like Yann LeCun reporting to him. Since joining, Wang has rejigged his team even more, which appears to have led to some discontent at the company.
Meta downplayed the departures, and said that Zhao had been slated to become the Chief AI Scientist from the outset. “We appreciate that there’s outsized interest in seemingly every minute detail of our AI efforts, no matter how inconsequential or mundane, but we’re just focused on doing the work to deliver personal superintelligence.” Meta told FT in a statement. “Some attrition is normal for any organisation of this size. Most of these employees had been with the company for years, and we wish them the best,” they added.
But it is unusual for employees to leave within weeks or months of joining a new company, as seems to have happened with a few Meta AI researchers. Meta’s Superintelligence Lab is an experiment at accelerating AI research — realizing that it was far behind other AI companies like Google, OpenAI or Anthropic, Meta appears to have looked to catch up by splurging massive sums of money on poaching some top researchers from rivals, and is looking to create cutting-edge AI models of its own. It remains to be seen how this experiment fares, but given the recent updates from the company, it does seem to be running into some early turbulence.