Thinking Machines co-founder Lilian Weng left the company citing health reasons, then joined OpenAI

Lilian Weng didn't mince words on her way out the door. In a Slack message she later posted to X, the Thinking Machines co-founder said the constant stress and workload had pushed her "beyond what my health can sustain physically." She'd apparently been wrestling with the decision for months before finally pulling the trigger this week.

Then came the twist. On Wednesday, OpenAI confirmed to TechCrunch that Weng is rejoining the company, this time to lead a top-level team focused on speeding up OpenAI's internal research. That team, notably, will touch work on recursive self-improvement — the idea of AI systems iterating on themselves to get more capable. Weng previously spent time at OpenAI as VP of AI Safety Research, so this isn't unfamiliar territory for her.

The optics are a little awkward. She left one job because the pace was crushing her, and landed at another lab racing on one of the most ambitious and pressure-filled problems in AI. The likely explanation is structural rather than contradictory: running a startup as a co-founder carries a different kind of weight than joining as a senior hire, even at a company moving just as fast on the technical side.

Mira Murati, the former OpenAI CTO who co-founded Thinking Machines with Weng, responded publicly to the exit, saying she was glad Weng was prioritizing her health and thanking her for their time building the company together. Whether Murati knew about the OpenAI move when she wrote that isn't clear.

What's clear is the signal this sends about the AI talent market right now. Departures at this level, especially from a company as closely watched as Thinking Machines, don't happen quietly. And when someone this senior resurfaces at a rival lab within days, it says something about how thin the bench of experienced AI safety and research leadership actually is — and how hard everyone is fighting to keep it.