Is there a mechanism in place to prevent malicious/fake reports?
Yes. While your overrides are immediately visible to you, changes are included in the publicly visible dataset only if they pass some trustworthiness criteria (including human validation).