Security in AI applied to healthcare
Prompt injection, jailbreaks, and improper behavior
Security prompts help assess risks from instruction manipulation, context leakage, improper responses, and protection failures in AI applications.
DR² ThinkTech
AI protection and security prompts
Prompts and guides to assess prompt injection, jailbreaks, data protection, and security risks in AI applied to healthcare.
Data protection
Risk, limits, and review before use
Security assessment should consider privacy, access control, logs, auditing, human review, and clear limits for use in clinical or corporate environments.