Paul Christiano is an AI alignment researcher and a central figure in the technical [[AI Safety]] community. He worked as a researcher at [[OpenAI]], where he made foundational contributions to reinforcement learning from human feedback (RLHF), the training technique that became critical to aligning large language models with human intent. His work on RLHF helped shape the methods underlying ChatGPT and similar systems. After leaving OpenAI, Christiano founded the Alignment Research Center (ARC), a nonprofit laboratory focused on developing techniques for evaluating and aligning advanced AI systems. ARC has conducted research on AI model evaluations, including assessments of whether AI systems can autonomously replicate or acquire resources, work that has influenced safety testing protocols across the industry. Christiano is widely regarded as one of the most technically rigorous voices in the alignment field. His research agenda emphasizes scalable oversight, iterated amplification, and eliciting latent knowledge from AI models. He has been a frequent participant in policy discussions about frontier AI risks and has contributed to shaping how leading AI laboratories approach safety benchmarks.