A verdict supports a decision. It does not replace one.
Authenticity examination is powerful, and like any tool used in consequential decisions, it can be misused. These are the principles we hold ourselves and our customers to.
Examination informs people; it does not judge them
A verdict is an input to a human decision, not an automatic conclusion about a person. We expect meaningful human review for consequential outcomes, especially where a result could affect someone’s access, livelihood, or reputation.
Calibrate thresholds to the stakes
No detector is perfect. Customers set the thresholds that determine when a result acts automatically and when it routes to a person. Higher-stakes decisions warrant more conservative thresholds and closer review. We provide the confidence score and per-signal breakdown precisely so these choices can be made deliberately.
Respect the people whose media is examined
Media often depicts real people. It should be handled with care, used only for the purpose it was collected, and retained no longer than necessary. Our security and data processing pages describe how we support this.
Uses we do not support
The platform is built for legitimate authenticity and fraud-prevention work. It is not intended for surveillance of individuals without a lawful basis, for harassment, or for any use that targets people on the basis of who they are. We reserve the right to decline or discontinue use that conflicts with these principles.
Questions or concerns
If you have a question about responsible use, or want to raise a concern about how the platform is being used, contact us at contact@safetrustlab.ai.