An authenticated agent can merge a conversation they control into a target conversation within a mailbox they are not authorized to access. This occurs because the `conversation merge` action records a failed authorization check in `response['msg']` but subsequently clears that value during the source-conversation loop. This flaw allows the injection of threads, attachments, and audit records into the unauthorized target and results in the deletion of the source conversation, although it does not disclose the existing contents of the target.