Lilian Weng, who helped start the AI company Thinking Machines Lab, has left the startup and gone back to work at OpenAI, according to TechCrunch. The news was reported on July 29, 2026. Weng had publicly said that the pressure and long hours at the startup had damaged her health.

Weng shared the news in a message to her coworkers at Thinking Machines and posted the same message on X. She wrote that the stress and workload had pushed her body past what it could handle. Business Insider reported that Weng had tried to rest and take sick leave while the company was rushing to release Inkling, its first product.

The situation is unusual. A co-founder leaves a startup worth $9 billion because of health problems, then quickly shows up at the company she had left before. OpenAI confirmed to TechCrunch that Weng would be rejoining. A spokesperson said she will lead a team focused on speeding up OpenAI's internal research, including work on recursive self-improvement. That is the idea of building AI systems that can get better at their own tasks on their own, without humans stepping in each time.

Weng is not a typical new hire for OpenAI. She previously worked there as VP of AI Safety Research — the team focused on making sure AI systems behave in ways that are safe and under control. She was also listed as a member of OpenAI's Safety and Security Committee with the title "Head of Safety Systems" when the committee was created in May 2024 (OpenAI). She left OpenAI in September 2024 (Bloomberg), then co-founded Thinking Machines Lab with Mira Murati, who had been OpenAI's Chief Technology Officer. The startup was founded in 2025 (Bloomberg). Thinking Machines has more than 30 people listed as members or employees, including notable researcher Andrew Tulloch (Bloomberg).

Murati replied publicly to Weng's X post, writing: "I'm glad that you're putting your health first. Thank you for everything." TechCrunch reported that it is unclear whether Murati knew Weng would rejoin OpenAI when she wrote that.

That question is worth pulling on. Weng said she left for health reasons, and she was direct that the cause was the demanding conditions of startup life. Whether OpenAI's environment will be any healthier is an open question. The irony of leaving a startup for health reasons and joining a company also known for its intensity is hard to ignore. But OpenAI's new assignment for Weng suggests this is not simply going back to her old job. The research area she will support — recursive self-improvement — sits right at the boundary between making AI more capable and keeping it safe. It is one of the most important unsolved problems in the field. Putting someone with Weng's safety background in a role designed to speed up that work suggests OpenAI treats safety expertise and pushing capabilities forward as two sides of the same coin, not opposing goals.

For Thinking Machines, losing a co-founder while racing to release its first product is a real loss. The company started in 2025 with a $9 billion valuation and a team of senior researchers. Losing a co-founder to burnout during a product rush, whatever happens next with staffing, tells us something about the human cost of the current AI startup boom. Murati handled the public departure gracefully, but what she knew about Weng's next move, and when she knew it, remains unanswered.

The broader context here is about how small and interconnected the top tier of AI research has become. Weng's path — from leading safety work at OpenAI, to co-founding a startup, and back to OpenAI in a role that blends safety with pushing AI capabilities further — says something about where the field is right now. The group of people working at the cutting edge of AI is small enough that the same names move between the major labs. And the line between safety research and capability research is getting harder to draw as AI systems become more advanced. Safety work used to act mostly as a brake on how fast companies pushed AI forward. Now, safety expertise is being built directly into that push.