Kai Yuanqing Xiao
Source-listed: Member of Technical Staff; o1 Safety Lead (project role) · OpenAI
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- IH-Challenge: A Training Dataset to Improve Instruction Hierarchy on Frontier LLMsSaved credit: Contributor · 2026Explore people connected to this work →
- Trading Inference-Time Compute for Adversarial RobustnessSaved credit: Contributor · 2025Explore people connected to this work →
- OpenAI o1 System CardSaved credit: Contributor · 2024Explore people connected to this work →
- The Instruction Hierarchy: Training LLMs to Prioritize Privileged InstructionsSaved credit: Contributor · 2024Explore people connected to this work →
- GPT-4 Technical ReportSaved credit: Contributor · 2023Explore people connected to this work →