AI Agent Safety Is Still Thinking Like Permissions(jeffreyflynt02.medium.com)5 points by jflynt76 33 days ago | 0 commentsNo comments yet