Use controls that match the risk of the tool
Different tools create different risks. MCXAI uses capability-specific controls instead of hiding every issue inside one generic warning.
Consent and identity
Voice, face and likeness workflows require explicit rights and consent. Impersonation, deception and non-consensual sexual content are prohibited.
Rights and transformation
Watermark, copyright and document transformation tools require authorization. Repeat abuse can lead to access restrictions.
High-stakes use
AI detector scores are not proof of misconduct; generated medical, legal, financial, educational and scientific output requires qualified review.