US, Allies Recommend Security Protections for AI Models

Cybersecurity Dive – Companies designing AI systems should protect training data from tampering and strictly limit access to its underlying infrastructure, the U.S.