The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions

In three linesOpenAI introduces an instruction hierarchy to train LLMs to prioritize privileged instructions and resist prompt injections and jailbreaks. The method enables models to distinguish system directives from malicious user inputs.

Summary generated by Claude — human-verified