viral-score

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow (“viral-score — is this clip worth publishing?”) ingests the user-provided clip/transcript and asks the model to score it against its rubric, meaning outsider-authored free text is directly read at runtime (e.g., “Here’s the transcript of a 2h podcast…” and other prompts).