Thinking Machines cofounders Barret Zoph and Luke Metz are leaving the fledgling AI lab and rejoining OpenAI, the ChatGPT-maker announced on Thursday. OpenAI’s CEO of applications, Fidji Simo, shared the news in a memo to staff Thursday afternoon.

The news was first reported on X by technology reporter Kylie Robison, who wrote that Zoph was fired for “unethical conduct.”

A source close to Thinking Machines said that Zoph had shared confidential company information with competitors. WIRED was unable to verify this information with Zoph, who did not immediately respond to WIRED’s request for comment.

Zoph told Thinking Machines CEO Mira Murati on Monday he was considering leaving, then was fired today, according to the memo from Simo. She goes on to write that OpenAI doesn’t share the same concerns about Zoph as Murati.

The personnel shake-up is a major win for OpenAI, which recently lost its VP of research, Jerry Tworek.

Another Thinking Machines Lab staffer, Sam Schoenholz, is also rejoining OpenAI, the source said.

Zoph and Metz left OpenAI in late 2024 to start Thinking Machines with Murati, who had been the ChatGPT-maker’s chief technology officer.

This is a developing story. Please check back for updates.

Since Donald Trump returned to the White House last January, the biggest names in tech have mostly fallen in line with the new regime, attending dinners with officials, heaping praise upon the administration, presenting the president with lavish gifts, and pleading for Trump’s permission to sell their products to China. It’s been mostly business as usual for Silicon Valley over the past year, even as the administration ignored a wide range of constitutional norms and attempted to slap arbitrary fees on everything from chip exports to worker visas for high-skilled immigrants employed by tech firms.

But after an ICE agent shot and killed an unarmed US citizen, Renee Nicole Good, in broad daylight in Minneapolis last week, a number of tech leaders have begun publicly speaking out about the Trump administration’s tactics. This includes prominent researchers at Google and Anthropic, who have denounced the killing as calloused and immoral. The most wealthy and powerful tech CEOs are still staying silent as ICE floods America’s streets, but now some researchers and engineers working for them have chosen to break rank.

More than 150 tech workers have so far signed a petition asking for their company CEOs to call the White House, demand that ICE leave US cities, and speak out publicly against the agency’s recent violence. Anne Diemer, a human resources consultant and former Stripe employee who organized the petition, says that workers at Meta, Google, Amazon, OpenAI, TikTok, Spotify, Salesforce, Linkedin, and Rippling are among those who have signed. The group plans to make the list public once they reach 200 signatories.

“I think so many tech folks have felt like they can’t speak up,” Diemer told WIRED. “I want tech leaders to call the country’s leaders and condemn ICE’s actions, but even if this helps people find their people and take a small part in fighting fascism, then that’s cool, too.”

Nikhil Thorat, an engineer at Anthropic, said in a lengthy post on X that Good’s killing had “stirred something” in him. “A mother was gunned down in the street by ICE, and the government doesn’t even have the decency to perform a scripted condolence,” he wrote. Thorat added that the moral foundation of modern society is “infected, and is festering,” and the country is living through a “cosplay” of Nazi Germany, a time when people also stayed silent out of fear.

Jonathan Frankle, chief AI scientist at Databricks, added a “+1” to Thorat’s post. Shrisha Radhakrishna, chief technology and chief product officer of real estate platform Opendoor, replied that what happened to Good is “not normal. It’s immoral. The speed at which the administration is moving to dehumanize a mother is terrifying.” Other users who identified themselves as employees at OpenAI and Anthropic also responded in support of Thorat.

Shortly after Good was shot, Jeff Dean, an early Google employee and University of Minnesota graduate who is now the chief scientist at Google DeepMind and Google Research, began re-sharing posts with his 400,000 X followers criticizing the Trump administration’s immigration tactics, including one outlining circumstances in which deadly force isn’t justified for police officers interacting with moving vehicles.

He then weighed in himself. “This is completely not okay, and we can’t become numb to repeated instances of illegal and unconstitutional action by government agencies,” Dean wrote in an X post on January 10. “The recent days have been horrific.” He linked to a video of a teenager—identified as a US citizen—being violently arrested at a Target in Richfield, Minnesota.

In response to US Vice President JD Vance’s assertion on X that Good was trying to run over the ICE agent with her vehicle, Aaron Levie, the CEO of the cloud storage company Box, replied, “Why is he shooting after he’s fully out of harm’s way (2nd and 3rd shot)? Why doesn’t he just move away from the vehicle instead of standing in front of it?” He added a screenshot of a Justice Department webpage outlining best practices for law enforcement officers interacting with suspects in moving vehicles.

How does procrastination arise? The reason you decide to postpone household chores and spend your time browsing social media could be explained by the workings of a brain circuit. Recent research has identified a neural connection responsible for delaying the start of activities associated with unpleasant experiences, even when these activities offer a clear reward.

The study, led by Ken-ichi Amemori, a neuroscientist at Kyoto University, aimed to analyze the brain mechanisms that reduce motivation to act when a task involves stress, punishment, or discomfort. To do this, the researchers designed an experiment with monkeys, a widely used model for understanding decisionmaking and motivation processes in the brain.

The scientists worked with two macaques that were trained to perform various decisionmaking tasks. In the first phase of the experiment, after a period of water restriction, the animals could activate one of two levers that released different amounts of liquid; one option offered a smaller reward and the other a larger one. This exercise allowed them to evaluate how the value of the reward influences the willingness to perform an action.

In a later stage, the experimental design incorporated an unpleasant element. The monkeys were given the choice of drinking a moderate amount of water without negative consequences or drinking a larger amount on the condition of receiving a direct blast of air in the face. Although the reward was greater in the second option, it involved an uncomfortable experience.

As the researchers anticipated, the macaques’ motivation to complete the task and access the water decreased considerably when the aversive stimulus was introduced. This behavior allowed them to identify a brain circuit that acts as a brake on motivation in the face of anticipated adverse situations. In particular, the connection between the ventral striatum and the ventral pallidum, two structures located in the basal ganglia of the brain, known for their role in regulating pleasure, motivation, and reward systems, was observed to be involved.

The neural analysis revealed that when the brain anticipates an unpleasant event or potential punishment, the ventral striatum is activated and sends an inhibitory signal to the ventral pallidum, which is normally responsible for driving the intention to perform an action. In other words, this communication reduces the impulse to act when the task is associated with a negative experience.

The Brain Connection Behind Procrastination

To investigate the specific role of this connection, as described in the study published in the journal Current Biology, researchers used a chemogenetic technique that, through the administration of a specialized drug, temporarily disrupted communication between the two brain regions. By doing so, the monkeys regained the motivation to initiate tasks, even in those tests that involved blowing air.

Notably, the inhibitory substance produced no change in trials where reward was not accompanied by punishment. This result suggests that the EV-PV circuit does not regulate motivation in a general way, but rather is specifically activated to suppress it when there is an expectation of discomfort. In this sense, apathy toward unpleasant tasks appears to develop gradually as communication between these two regions intensifies.

Beyond explaining why people tend to unconsciously resist starting household chores or uncomfortable obligations, the findings have relevant implications for understanding disorders such as depression or schizophrenia, in which patients often experience a significant loss of the drive to act.

However, Amemori emphasizes that this circuit serves an essential protective function. “Overworking is very dangerous. This circuit protects us from burnout,” he said in comments reported by Nature. Therefore, he cautions that any attempt to externally modify this neural mechanism must be approached with care, as further research is needed to avoid interfering with the brain’s natural protective processes.

This story originally appeared in WIRED en Español and has been translated from Spanish.