OpenAI Explains Prompt Injection: A Frontier Safety Challenge for AI Systems
Understanding prompt injections: a frontier security challenge
AI summary
OpenAI views prompt injection as a frontier safety challenge for AI systems and explains how such attacks work. OpenAI says it is advancing related research, strengthening protections in trained models, and building guardrails for users.