The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions

Company news OpenAI Source language: English

Today's LLMs are susceptible to prompt injections, jailbreaks, and other attacks that allow adversaries to overwrite a model's original instructions with their own malicious prompts.

Excerpt

Today's LLMs are susceptible to prompt injections, jailbreaks, and other attacks that allow adversaries to overwrite a model's original instructions with their own malicious prompts.

Short excerpt from the source — open the original for the full text.

Sources

← Back to the daily brief