hr-employer-branding

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text is likely ingested because the required workflow explicitly instructs the agent to “read the negative reviews in full” and “talk directly to a sample of recent declined candidates,” which would pull in third-party review/comment text and candidate-provided free-form reasons into the LLM context.