Why AI Pioneer Lilian Weng Quit Her Unicorn for Health: A Candid Look

On July 28, Lilian Weng (Lilian Weng), a renowned AI safety researcher and former OpenAI senior leader, announced on X that the next day would be her last at Thinking Machines Lab, citing severe health problems that had plagued her for seven months.

The article outlines her background: born in Shaanxi, she won a national math competition in 2004, studied information management at Peking University, earned a PhD at Indiana University focusing on network diffusion and complex systems, joined OpenAI in 2018, contributed to the Dactyl robotic‑hand project, built the Safety Systems team to over 80 members, and was promoted to Vice President of Research & Safety before co‑founding Thinking Machines Lab in February 2025.

Thinking Machines Lab quickly achieved milestones: it released the open‑source model Inkling, secured a $2 billion seed round led by a16z with Nvidia, AMD and Jane Street participating, and reached a post‑money valuation of $12 billion. Apple and Meta later offered investment or acquisition, which co‑founder Mira Murati declined. However, the company’s ambitious plan for a $50‑60 billion financing round collapsed in early 2026, causing the valuation to revert to the original $12 billion, and several key founders and researchers departed.

The piece places Weng’s departure in the broader AI industry context, describing how rapid model releases, escalating compute costs, and a four‑front race on product, compute, funding, and talent create relentless pressure. She emphasizes that while AI systems can scale indefinitely, human bodies cannot, and her decision reflects a need to align personal limits with the industry’s relentless speed.

Reactions from the community were supportive: co‑founder Mira Murati praised Weng’s choice, and AI leaders such as Soumith Chintala and Saurabh Garg echoed that health must come first. Weng’s farewell letter contains no discussion of financing or internal conflict; it solely underscores the core tension between human endurance and the ever‑accelerating AI race.