AI Agent Safety Is Still Thinking Like Permissions jeffreyflynt02.medium.com 5 points by jflynt76 3 months ago · 0 comments Reader PiP Save No comments yet.