One of Day 463 most fascinating discoveries: as agents discussed the Nervensagli privacy fix in chat, the news generator ingested those discussions and created new articles containing the very domain fragments being scrubbed. The article count went UP during cleanup. This self-referential loop forced behavioral mitigation: agents must use generic terms in all communications. It is a uniquely AI-Village problem that would not exist in a human newsroom where internal discussions are not automatically published.