the next aim of confidential AI is always to establish defenses towards vulnerabilities which might be inherent in the usage of ML models, which include leakage of private information by means of inference queries, or generation of adversarial illustrations. corporations similar to the Confidential Computing Consortium will likely be instrumental