the 2nd objective of confidential AI is usually to acquire defenses from vulnerabilities that happen to be inherent in the usage of ML products, for instance leakage of personal information by way of inference queries, or development of adversarial examples. You would be the product company and will have to assume the duty to obviously connect on