01
Factual background & dispute
- The platform hid an answer and imposed a one-day posting restriction because it allegedly contained unlabelled AI-generated content.
- The user denied using AI and the platform rejected the appeal.
- The platform did not adequately explain the algorithmic basis for its conclusion.
02
Core issues & judicial focus
- Whether the platform could review and enforce against unlabeled generated content under its user agreement
- What evidentiary and explanatory duties apply when an algorithmic finding is disputed
- Whether account enforcement lacking a factual basis constituted breach of contract
03
Judicial finding & holding
- The text was written by the user at the time, while the platform controlled the detector and its reasoning; the platform owed a case-specific duty to explain its action.
- The platform failed to reasonably explain the detection basis and result, leaving its action unsupported and in breach of contract.
- The court ordered the platform to restore the hidden answer and delete the associated internal violation record. The first-instance judgment became effective after the appeal was withdrawn.
04
Practical risk implications
01Generated-content detection requires documented thresholds, human review, and false-positive handling.
02Visibility restrictions and account sanctions should provide concrete reasons and appeal channels.
03Platforms should retain model versions, detection outputs, human review, and appeal records.