Inside OpenAI’s raid on the Thinking Machines Lab

If someone ever He’s making an HBO Max series about the AI ​​industry, and this week’s events will make for a great episode.

On Wednesday, Fidji Simo, chief applications officer at OpenAI, announced that the company had rehired Barret Zoph and Luke Metz, co-founders of AI startup Thinking Machines Lab founded by Mira Murati. Zoff and Metz left OpenAI in late 2024.

Last night we reported on two narratives forming around the reasons behind the departure, and we have since learned new information.

A source with direct knowledge says Thinking Machines leadership believes Zoph was involved in a serious misconduct incident while at the company last year. The source says that this incident shattered Moratti’s confidence and disrupted their working relationship. The source also claimed that Moratti fired Zoff on Wednesday — before he knew he was going to OpenAI — over what the company claimed were issues that arose after alleged misconduct. At the time the company learned of Zoph’s return to OpenAI, Thinking Machines raised concerns internally about whether he had shared confidential information with competitors. (Zoph did not respond to several requests for comment from WIRED.)

Meanwhile, in a Wednesday memo to employees, Simo claimed that employees had been on the job for weeks and that Zoff told Moratti he was considering leaving Thinking Machines on Monday — before his dismissal date. Simo also told employees that OpenAI does not share Thinking Machines’ concerns about Zoph’s ethics.

Along with Zoph and Metz, Sam Schoenholz, another former OpenAI researcher who worked on Thinking Machines, is joining the maker of ChatGPT, according to Simo’s announcement. At least two Thinking Machines employees are expected to join OpenAI in the coming weeks, according to a source familiar with the matter. Technology reporter Alex Heath was first to report the additional appointments.

A separate source familiar with the matter dismissed the perception that the recent personnel changes were entirely related to Zoph. “This was part of a long discussion at Thinking Machines. There was discussion and back-and-forth about what the company wanted to build, and it was about the product, the technology, and the future.”

Thinking Machines Lab and OpenAI declined to comment.

In the wake of these events, we’ve heard from many researchers at leading AI labs who say they’re exhausted by the ongoing drama in their industry. This specific incident is reminiscent of OpenAI’s brief ouster of Sam Altman in 2023, known within OpenAI as the “blip.” Moratti played a key role in the event as the company’s chief technology officer at the time, the Wall Street Journal reports.

In the years since Altman’s ouster, drama in the AI ​​industry has continued, with the departure of co-founders of several major AI labs, including xAI’s Igor Babushkin, Safe Superintelligence’s Daniel Gross, and Meta’s Yann LeCun (co-founder of Facebook’s long-running AI lab, FAIR).

Some might argue that this drama is justified for a nascent industry whose expenses contribute to America’s GDP growth. Also, if you believe in the idea that one of these researchers might make some breakthroughs on the path to artificial general intelligence, it’s worth keeping track of where they’re headed.

However, many researchers began working before ChatGPT’s success and seem surprised that their industry is now a source of almost constant scrutiny.

As long as researchers can continue to raise $1 billion seed rounds on a whim, we’re guessing that the power changes in the AI ​​industry will continue apace. HBO Max writers, shut the door.

How AI labs train agents to do your job

People in Silicon Valley have been thinking that AI will replace jobs for decades. However, in the past few months, efforts to make AI do work of economic value have become much more complex.

AI labs improve the data they use to create AI agents. Last week, WIRED reported that OpenAI asked third-party contractors from Handshake to upload examples of their real work from previous jobs to evaluate OpenAI agents. Companies require employees to clean these documents of any confidential data and personally identifiable information. While it’s possible that some company secrets or names could escape, that’s likely not what OpenAI is going for (although the company could be in serious trouble if that happened, experts say).