AI SAFETY VIA DEBATE - AN OVERVIEW

The second goal of confidential AI will be to develop defenses versus vulnerabilities which have been inherent in the usage of ML versions, like leakage of personal information via inference queries, or generation of adversarial examples. We endorse which you have interaction your authorized counsel early inside your AI challenge to evaluate your