designing-ai-agents-to-resist-prompt-injection.log
|src: openai.com
Designing AI agents to resist prompt injection
How ChatGPT defends against prompt injection and social engineering by constraining risky actions and protecting sensitive data in agent workflows.
