c. Incorporate downstream measures and processes for users or stakeholders to verify how and when AI technology is being applied.

AI systems should be safe and secure, not vulnerable to tampering or compromising the data they are trained on.