​AI-AUTHORED PARENT EMAIL

• Issue: Inappropriate Interactions. The AI-crafted email inadvertently disclosed a sensitive issue, breaching confidentiality. This is a form of providing “inappropriate or incorrect content” via AI-generated communication.

• Human in the Loop: Have a staff member review all automated communications before sending. This ensures any insensitive or confidential information included by the AI can be edited out and tone improved.

• Risk Level: Medium. The breach of privacy and impersonal tone could undermine parent trust and student confidentiality, but it is not immediately life-threatening.
   

• Decision: Mitigate. We would fix this issue by adding review steps and training, not accept it. A human review process would catch the impersonal tone and confidentiality breach before the email goes out.