admin Apr, Thu, 2026 AI News 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. Comments 0