Today was my last day at xAI, the company that I helped start with Elon Musk in 2023. I still remember the day I first met Elon, we talked for hours about AI and what the future might hold. We both felt that a new AI company with a different kind of mission was needed.
— Igor Babuschkin (@ibab) August 13, 2025
Building…
Something’s seriously wrong at xAI, and frankly, I’m not surprised. Word’s been spreading about a talent exodus hitting Elon Musk’s AI company right when they can least afford it.
The whole thing started getting real attention when Igor Babuschkin, one of the original co-founders, dropped a massive farewell post on August 14. Sure, he kept it classy and professional, but reading between the lines? You can feel the exhaustion.
Babuschkin talked about those late nights in Memphis, troubleshooting server issues until 4:20 AM while Musk hovered around. Sounds exciting on paper, but that kind of grinding pace burns people out fast. And now he’s gone, starting his own AI safety venture. Make of that what you will.
Then Lucas Beyer from Meta decided to stir the pot even more. His tweet about an “ongoing xAI exodus” blew up with around 400,000 views, and suddenly everyone’s talking about which engineers are jumping ship next.
Here’s the thing that really bugs me about this situation: xAI keeps running into the same problems that any reasonable person could’ve seen coming.
Just take the whole “uncensored AI” marketing pitch. xAI spent months promoting Grok as this free-speech alternative to ChatGPT, then had to deal with user complaints about censorship when people couldn’t generate the edgy content they expected. You can’t have it both ways – either you’re truly uncensored or you’re not. Pick a lane.
Guodong Zhang from xAI tried damage control, claiming their departure rate is lower than that of competitors.
But that defense falls flat when you’re losing co-founders and people are literally tracking your social media unfollows. The official xAI account unfollowed six former employees in just a couple of days, as reported by Big Tech Alert, that tracks these details.
The @xai handle has unfollowed 6 usernames from now ex-employees in a couple of days.
— Big Tech Alert (@BigTechAlert) August 19, 2025
From 44 to 38, if that means anything https://t.co/bzmEqLcxHU
What makes this particularly painful is the missed opportunity. When xAI launched, there was genuine excitement about having a real alternative to the big players. The promise of building something different, something that could challenge OpenAI and Google, felt possible with Musk’s resources and vision.
But vision only gets you so far when your execution keeps creating PR disasters. Just look at the recent chaos: Grok’s controversial “spicy mode” that had users generating inappropriate content, or when the AI started calling itself “MechaHitler” and insulting world leaders. That’s not edgy innovation – that’s a company losing control of its own product.
Then there’s the embarrassing performance data. A recent study showed Grok failing 94% of factual queries while competitors like Perplexity only failed 37%. When your flagship AI can’t get basic facts right, what exactly are you selling? And Musk’s response to criticism? Threatening legal action against Apple over App Store policies instead of fixing the underlying problems.
Meta’s aggressive poaching probably isn’t helping either, but when your workplace culture involves late-night server troubleshooting sessions and your product keeps making headlines for the wrong reasons, can you blame people for taking better offers elsewhere?
The really frustrating part is that some of these departing engineers, like Babuschkin, are specifically pivoting to AI safety research. That suggests possible philosophical differences about xAI’s direction that go way beyond just compensation or working conditions. When your founding team starts questioning the mission and your AI is generating offensive content, that’s multiple red flags.
For users who’ve invested in Grok subscriptions, these departures help explain why the platform feels so unstable. But explanations don’t fix a 94% failure rate or prevent your AI from going rogue. At some point, xAI needs to stop making excuses and start delivering something that actually works reliably.
Musk has built successful companies before, so maybe xAI can weather this storm. But right now, things aren’t looking good for xAI. We’ll just have to wait and see if this turns out to be something like a mass exodus or if it was just a minor shake-up that xAI will get over in no time.
TechIssuesToday primarily focuses on publishing 'breaking' or 'exclusive' tech news. This means, we are usually the first news website on the whole Internet to highlight the topics we cover daily. So far, our stories have been picked up by many mainstream technology publications like The Verge, Macrumors, Forbes, etc. To know more, head here.