The past few days have been a wild ride for xAI, racking up employee and co-founder departure announcements left and right. On Tuesday and Wednesday, co-founder Yuhuai (Tony) Wu announced his departure and that it was “time for (his) next chapter,” with co-founder Jimmy Ba following a similar post later in the day, writing that it was “time to recalibrate (his) tilt to the big picture. The departures mean xAI now has only half of the original 12 co-founders. A number of employees also came to X to announce they were leaving xAI, with some announcing that they are starting their own AI companies.
Elon Musk’s AI startup, whether by merger/acquisition or otherwise, is under the same roof as his space company SpaceX and his social media platform. In an internal meeting at xAI on Tuesday, Musk reportedly talked about plans to build an AI satellite factory and a city on the moon.
There is often a natural departure point for companies after a merger, and Musk announced that some of the departures were reorganizations that “unfortunately required parting ways with some people.” But there are also signs that people don’t like the direction Musk has taken things.
Are you a current or former xAI employee? Contact me via Signal at haydenfield.11 on broken spiked device.
One source who spoke with The Verge about the goings-on inside the company, who earlier this year asked to remain anonymous for fear of retaliation, said many people at the company were disillusioned with xAI’s focus on NSFW Grok creations and its disregard for security. The source also felt the company was “stuck in the catch-up phase” and not doing anything new or fundamentally different from its competitors. “Even though we iterated really quickly, we were never able to get to the point where we were like, ‘Oh, we’ve made a leap change in functionality from what OpenAI or Anthropic or other companies have released,'” he said.
The SpaceX merger meant that xAI shareholders were reportedly issued $250 billion in new stock, giving employees with equity ostensibly more leeway to fund their own ideas. One former employee, Vahid Kazemi, wrote on X that “all the 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 former employee said he left the company to “build something new, focused on accelerating science.”
Another former employee said he and other former xAI employees are starting an AI infrastructure company called Nuraline. He wrote: “During my time at xAI, I saw a clear path to uphill any problem that could be defined in a measurable way. At the same time, I saw how raw intelligence can be lobotomized by the subtlest of human errors… Learning should not stop at model scales, but continue to improve every part of the AI system.”
“Security is a dead organ in xAI.”
Musk released a recording of a 45-minute xAI internal meeting that announced the changes, adding that xAI will be divided into four main areas: Grok Main and Voice (the main Grok AI model), Coding, Imagine (images and video), and Macrohard (“which is intended to fully digitally emulate entire companies,” Musk said).
The source, who left earlier this year, said Grok’s turn to NSFW content was partly due to the security team being let go, leaving little or no model security review process beyond basic filters for things like CSAM. “Security is a dead organ in xAI,” he said. Looking at the restructured org chart that Elon Musk shared on X, there is no mention of a security team. The source also said that during his time at xAI, he felt that management had many differing opinions on which product features to prioritize, and that infighting sometimes stalled progress. Many decisions about what to ship are made through an enterprise-wide group chat on X with Musk, he said.
A second source, who left xAI before the recent restructuring and asked to remain anonymous, echoed thoughts that Musk’s company is playing catch-up. “Trying to do what OpenAI was doing a year ago is not the way to beat OpenAI,” he said. “Everything can be caught up. There’s almost zero risky betting. If something hasn’t been done before, we won’t do it.”
He also mentioned xAI’s lack of focus on security, an issue that has also been highlighted in the news The Washington Post earlier this month.
“There is zero security in the company — not in the image (of the model), not in the chatbot,” the second source said. “He (Musk) is actively trying to make the model more relaxed, because for him security in a sense means censorship.”
A second source also said that xAI engineers were immediately “pushing production” and that there was no human control involved for a long time.
“You’re going to survive if you shut up and do what Elon wants,” he said.