Lilian Weng

Lilian Weng is the former Vice President of Research, Safety at OpenAI, where she played a pivotal role in advancing AI technology and safety during her seven-year tenure. Joining OpenAI in 2017, she initially contributed to robotics, leading the groundbreaking project to teach a robotic hand to solve a Rubik’s Cube. As a research scientist and technical lead, she specialized in algorithms for training complex robotic tasks.

‍

As OpenAI transitioned to large language models (LLMs), Weng helped shape their development and safety systems, pioneering research that set industry standards. She founded and led the Applied AI Research team, delivering foundational tools such as the fine-tuning API, embedding API, moderation endpoints, and applied safety frameworks—enhancing the power, utility, and safety of OpenAI’s APIs.

‍

Following the launch of GPT-4, Weng unified OpenAI’s safety efforts into a single team of over 80 scientists and engineers, driving key advancements in adversarial robustness, safety evaluations, and deployment infrastructure for building the full safety stack for AI development and deployment at OpenAI. Weng’s leadership of the unified Safety Systems team was recognized on Business Insider's 2024 AI Power List, given the team’s contribution to safeguard OpenAI’s frontier models and mitigate risks.

‍

Weng’s work has set benchmarks for AI safety, from balanced safety policies and improved model robustness to the development of OpenAI’s safest model to date—the o1-preview—which resists adversarial attacks while maintaining utility.

‍

She holds a bachelor's degree from Peking University and a Ph.D. from Indiana University Bloomington.

‍