One of the most respected figures in AI safety is making a surprising round-trip between two of the field's most watched companies.
Lilian Weng, a co-founder of Mira Murati's startup Thinking Machines, announced this week that she was stepping down, citing health reasons. In an internal Slack message she also shared publicly on X, Weng wrote: "I don't feel I'm able to continue at the pace a startup requires... the amount of consistent stress and workload have pushed me beyond what my health can sustain physically."Within days, OpenAI confirmed to TechCrunch that Weng is rejoining the company — where she previously served as VP of AI Safety Research. According to an OpenAI spokesperson, she will lead a top-level team focused on accelerating the company's internal research, with a particular emphasis on recursive self-improvement — the idea that an AI system could iterate on itself to become progressively more capable.
The move is notable on two fronts. First, it's a high-profile defection from Thinking Machines, one of the best-funded and most closely followed new AI labs, just as it scales. Second, it places Weng — long a voice for caution — at the heart of a research direction (recursive self-improvement) that many consider among the most powerful and potentially risk-laden pathways toward more advanced AI.
The news was first reported by The Information and expanded on by TechCrunch and Business Insider. Weng framed her departure as a personal health decision rather than a verdict on Thinking Machines' trajectory, though the juxtaposition of leaving a startup for its punishing pace and immediately joining an even larger one drew attention across the industry.
Read more on our ongoing AI talent coverage.
