
OpenAI Blog··ongeveer 1 maand geleden
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.
Lees origineel artikel



