content-research-and-sourcing

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

• Third-party content exposure detected (high risk: 0.75). The required workflow is a verification/citation protocol that explicitly involves opening and checking AI-supplied or inherited citations (i.e., runtime web content/links that may contain outsider-authored free text), which the agent would read to verify “SAYS IT” and “LOG IT.”