The AI drama is heating up

Everywhere you look in the world of AI, knives are coming out.

Take the recent bickering between Elon Musk and Yann LeCun on X, where Musk’s rivals in the AI race seemingly can’t stop feeding him more training data. This particular spat started with LeCun, Meta’s top AI researcher, criticizing Musk for spewing “crazy-ass conspiracy theories” and pointing out that anyone who joins his OpenAI competitor, xAI, will have to take orders from an erratic megalomaniac.

LeCun was responding to Musk’s touting of xAI’s record-breaking, $6 billion funding round, news that Musk shared with the call for more researchers to apply for open roles. While I have no doubt that LeCun has a personal distaste for Musk, their exchange is emblematic of a broader trend that’s playing out: as the stakes in the AI race get higher and the fight for talent grows fiercer, the drama is heating up.

The announcement of xAI’s war chest this week will only create more tension across the other AI labs as Musk continues his poaching spree. Within Meta, I’ve heard there’s deep insecurity about losing more senior researchers after a string of recent departures. There’s also the report that both Meta and xAI are pursuing some kind of partnership with the buzzy startup Character.ai, which I’m told has a new foundational model in the works and will likely raise more money soon.

Despite its outsize and growing importance, the AI talent pool remains incredibly small. It’s a great situation for the handful of rockstar researchers out there. It puts more pressure than ever on leaders like Musk and LeCun to attract the best people.

“You can disagree, but there’s maybe 5-7 ML researchers in the world that can attract enough talent to build a large industrial lab,” Arun Rao, a product manager in Meta’s genAI group, recently wrote on Threads. “The only ones who’ve done it are Ng, Yann, Hinton, Ilya, & maybe Dario… Note that there are dozens of superstar researchers and thousands of good ones - none can attract talent and capital like this small group.”

With growing power concentrated in such a small group, it’s only natural for egos to collide. Feuds are also being animated by the ungodly amount of money being pumped into the ecosystem. Investors have now valued xAI at $24 billion — a ridiculous number (which I first reported in a previous issue of this newsletter) that only makes sense because of Elon Musk.

Most of this $6 billion round is going to pay for GPUs owned by Oracle. If the last couple of years are any indication, it won’t be nearly enough. I won’t be surprised if Musk raises billions of more dollars for xAI within a year.

The most dramatic company in AI is of course OpenAI, where we’re starting to finally hear more details about last year’s failed coup attempt against Sam Altman.

On a recent podcast, ex-board member Helen Toner accused Altman of “lying to other board members in order to try and push me off the board” before she moved to fire him alongside the now-defenestrated Ilya Sutskever. She said Altman didn’t disclose that he owned OpenAI’s VC fund and that the board learned about the launch of ChatGPT on Twitter — admissions that are more embarrassing for the board than Altman.

According to Toner, Altman gave the previous board inaccurate information about the company’s safety processes. (He is now one of four board members on a special safety committee.) In addition, she confirmed earlier reports that the board was approached by unnamed OpenAI execs accusing Altman of “psychological abuse” and “lying and being manipulative in different situations.”

Toner and fellow ex-board member Tasha McCauley are clearly on some kind of revenge tour. They also fired shots in a recent op-ed in The Economist: “The board’s ability to uphold the company’s mission had become increasingly constrained due to long-standing patterns of behavior exhibited by Mr Altman, which, among other things, we believe undermined the board’s oversight of key decisions and internal safety protocols.”

Their missive prompted a response op-ed from current OpenAI board members Bret Taylor and Larry Summers who said they “do not accept the claims made by Ms Toner and Ms McCauley.” They went on to repeat that OpenAI conducted an independent investigation clearing Altman and has since strengthened company safety protocols and the conflict of interest policy.

The investigation itself remains under lock and key alongside the specifics of OpenAI’s updated policies. (This is probably where I should disclose that my employer, Vox Media, has announced a content licensing deal with OpenAI — it has no impact on my coverage.) With OpenAI loosening its restrictions on what departing employees can say publically about their experience, and some already choosing to speak out about their concerns, I suspect that drama is only the beginning.

My notes on what else is happening right now:

Some notable career moves I’ve seen recently:

If you aren’t already subscribed to Command Line, don’t forget to sign up and get future issues delivered directly to your inbox. You’ll also get full access to the archive featuring scoops on companies like Meta, Google, X, OpenAI, and more.

As always, I want to hear from you, especially if you have a juicy tip or piece of AI drama. You can ping me securely on Signal. If you want to talk about something sensitive, I’d recommend not using a work device. I’m happy to keep you anonymous.

Thanks for subscribing.