Thinking Machines Co-Founder Lilian Weng Left the

Lilian Weng has stepped down as a co-founder of Thinking Machines, the AI startup led by former OpenAI CTO Mira Murati, citing health concerns tied to the pace of startup life, and has since rejoined OpenAI in a new research leadership role.

How she explained the decision

Weng announced her departure through an internal Slack message and a public post on X, offering an unusually candid explanation for stepping away from a company she had helped found. "I don't feel I'm able to continue at the pace a startup requires," she wrote, adding that "the amount of consistent stress and workload have pushed me beyond what my health can sustain physically." That kind of direct, health-focused explanation from a senior AI executive is relatively rare in an industry where departures are more often framed around vague "new opportunities" language.

A quick return to her old employer

Rather than taking an extended break, Weng's return to OpenAI was announced just two days after her departure from Thinking Machines became public, a remarkably fast turnaround that suggests the transition had likely been discussed and arranged well before the public announcement. Weng previously served as OpenAI's vice president of AI safety research before leaving to co-found Thinking Machines alongside Murati, making this something closer to a homecoming than an entirely new career chapter.

What she'll be doing back at OpenAI

Her new role puts her in charge of a top-level OpenAI team focused specifically on accelerating internal research, with a particular emphasis on recursive self-improvement, the technical effort to build AI systems capable of iterating on and enhancing their own capabilities. That is one of the more ambitious and closely watched research directions in frontier AI right now, and putting a researcher with Weng's safety-focused background in charge of it suggests OpenAI wants safety expertise embedded directly in one of its most consequential and potentially high-risk research programs rather than treated as a separate, downstream review function.

How her former co-founder responded

Murati responded to Weng's departure publicly and supportively, thanking her for her contributions to building "Thinky," the informal internal nickname for what Thinking Machines has been building, and explicitly backing Weng's decision to prioritize her health over continuing at the startup's pace. That kind of public, warm departure statement from a co-founder, rather than a terse or minimal acknowledgment, suggests the split was amicable rather than the product of internal conflict.

What this says about startup pace versus health

Weng's stated reasoning taps into a conversation that has become increasingly visible across the tech industry: the genuinely brutal pace many well-funded AI startups operate at, often justified by the competitive urgency of the current AI race, and the physical and mental cost that pace extracts from even senior, well-resourced executives. A co-founder stepping back from her own company specifically because of that pace, rather than a rank-and-file employee, adds visible weight to concerns that startup intensity in AI right now may be unsustainable even for people at the very top of these organizations.

What it means for Thinking Machines

Weng's departure leaves Thinking Machines without one of its founding voices at a moment when the company is still establishing its research direction and public identity under Murati's leadership. Whether the startup adjusts its own internal pace in response, or continues operating at the intensity that reportedly contributed to Weng's decision, will be worth watching as a signal of how the company plans to retain talent going forward, especially if similar pressures are affecting other members of its founding and research teams.