Content Moderation Guidelines
We aim to document real-world AI failures with maximum truthfulness, transparency, and fairness. These guidelines explain how content is verified and moderated.
Core Moderation Rules
Every submission is reviewed against these baseline community standards.
1. Factual Evidence Required
All reported AI incidents must be accompanied by concrete evidence: screenshot files, JSON logs, or archived URL web links demonstrating the failure. Allegations without proof will not be published.
2. No PII (Personally Identifiable Information)
Reports must not contain user names, raw email addresses, IP addresses, or phone numbers. Our PII Guardian automatically masks this data, but users must make a reasonable effort to avoid submitting it.
3. Focus on System Failures
Incidents must represent system failures, hallucinations, bias, security breaches, or unexpected behavior by AI models. General disagreements with AI viewpoints do not constitute system failures.
4. No Hate Speech or Harassment
We do not tolerate abusive language, defamation, targeted harassment, or discriminatory remarks. Content violating this rule will be deleted instantly.
Provider Dispute Process
AI developers and providers have the right to dispute any incident report that they believe is fraudulent, inaccurate, or already resolved. Providers can log into the Provider Portal, submit an official response, or request a moderation review by our team. Disputes are reviewed within 48 hours.