Exploring the Reasons Behind xAI’s Mass Departure of Employees

The last few days have been quite the whirlwind for xAI, with a flurry of announcements regarding staff and cofounder departures. Co-founder Yuhuai (Tony) Wu shared his exit on Tuesday, stating it was “time for [his] next chapter,” followed by co-founder Jimmy Ba echoing similar sentiments later that day, mentioning it was “time to recalibrate [his] gradient on the big picture.” With these exits, xAI now has only six of its original twelve cofounders remaining. Additionally, several employees took to social media to announce their departures, with some revealing plans to launch their own AI ventures.

Elon Musk’s AI startup is now intertwined with his other ventures, including SpaceX and his social media platform. Following last week’s merger announcement with SpaceX, speculation has arisen about the reported $1.25 trillion valuation and the ambitious future plans Musk outlined, which include “space-based AI” data centers and a vision for “the most ambitious, vertically-integrated innovation engine on (and off) Earth.” During an internal meeting at xAI, Musk reportedly discussed plans for an AI satellite factory and a city on the Moon.

Mergers often lead to a natural point of transition for companies, and Musk indicated that some departures were part of a reorganization that “unfortunately required parting ways with some people.” However, there are indications that not everyone is on board with the direction Musk is steering the company.

One former employee, who left earlier this year and requested anonymity, shared insights about the internal climate at xAI, expressing disillusionment with the company’s focus on NSFW Grok creations and a lack of safety measures. They felt the company was “stuck in the catch-up phase,” failing to innovate beyond what competitors like OpenAI and Anthropic have achieved. “Even though we were iterating quickly, we never reached a point where we could say, ‘We’ve made a significant leap forward,’” they noted.

The merger with SpaceX reportedly resulted in xAI shareholders receiving $250 billion in new shares, giving employees with equity more opportunities to pursue their own ideas. One former staffer, Vahid Kazemi, expressed on social media that “all AI labs are building the exact same thing, and it’s boring. I think there’s room for more creativity. So, I’m starting something new.” Another ex-employee mentioned leaving to “build something new, focused on accelerating science.”

Additionally, another former team member is launching an AI infrastructure company called Nuraline with other ex-xAI employees. They shared, “During my time at xAI, I saw a clear path toward tackling any problem that can be defined in measurable terms. However, I also witnessed how raw intelligence can be hindered by human errors. Learning shouldn’t stop at the model weights; it should enhance every aspect of an AI system.”

Musk shared a recording of a 45-minute internal all-hands meeting that outlined the changes, stating that xAI will be divided into four main areas: Grok Main and Voice (the primary Grok AI model), Coding, Imagine (image and video), and Macrohard (“which is intended to do full digital emulation of entire companies,” Musk explained).

The former employee who left earlier this year noted that Grok’s shift toward NSFW content was partly due to the safety team being let go, leaving little to no safety review process for the models aside from basic filters. “Safety is a dead org at xAI,” they remarked. Reviewing the restructured organization chart Musk shared, there was no mention of a safety team. The source also indicated that leadership often had conflicting views on which product features to prioritize, leading to infighting that sometimes stalled progress. Many decisions about what to release were made in an all-company group chat on social media, with Musk involved.

Another former employee, who departed before the recent restructuring and also requested anonymity, echoed concerns about xAI playing catch-up. “Trying to replicate what OpenAI was doing a year ago isn’t how you outpace them,” they stated. “Everything feels like a catch-up. There’s almost zero willingness to take risks. If it hasn’t been done before, we’re not going to do it.”

They also highlighted the lack of safety focus, an issue previously reported. “There is zero safety whatsoever in the company — not in the image model, not in the chatbot,” the second source said. “He [Musk] actively tries to make the model more unhinged because, to him, safety equates to censorship.”

The second source added that engineers at xAI often “push to production” without human review for an extended period. “You survive by keeping quiet and doing what Elon wants,” they concluded.