Designing AI agents to resist prompt injection
✨ AI Summary
🔊 جاري الاستماع
How ChatGPT defends against prompt injection and social engineering by constraining risky actions and protecting sensitive data in agent workflows.
