New Norms for AI: Zero Trust—Verify Then Trust

Abstract

This conceptual article highlights an important puzzle concerning the actions of prominent AI industry leaders calling for a six-month halt to the further development and training of AI, while some have already deployed and continue to deploy AI; a so-called arms race. As security risks increase at scale, managing the security of AI is an essential consideration. Although implementing standards for AI is currently a preferred solution for many, this article argues that, on their own, standards may not provide sufficient capability and flexibility. An argument is presented that cyber norms of zero trust for AI (ZTAI) alongside standards are essential for the security of AI, along with a call to action.