Pamela Mishkin is a Member of Technical Staff at OpenAI, where she focuses on the policy aspects of large language model deployment, working at the intersection of technical research and AI policy to make models safer and fairer.
Mishkin is a co-author on several of OpenAI's most influential papers, including "Training language models to follow instructions with human feedback" (InstructGPT), the foundational work that shaped ChatGPT's use of reinforcement learning from human feedback; "Learning Transferable Visual Models From Natural Language Supervision" (CLIP); and the Codex paper underlying GitHub Copilot. More recently, she co-authored OpenAI's 2025 white paper "OpenAI's Approach to External Red Teaming for AI Models and Systems," describing the company's methodology for stress-testing AI systems to surface risks ahead of deployment.
Mishkin holds a degree from the University of Cambridge and began at OpenAI as an OpenAI Scholar in 2020, working on adversarial perturbations in language models.
