At the Secure Minds Summit in Las Vegas, experts discussed the dual potential and risks of AI in security and application development, particularly focusing on the importance of permissions and access controls. AI agents, if not properly managed, can inadvertently access sensitive data, leading to potential chaos within organizations. The need for robust permissions systems is crucial to ensure AI agents only access data they are authorized to use, aligning with organizational policies. AuthZed has developed an Authorization Infrastructure for AI designed to respect permissions, prevent data leaks, and provide audit trails, powered by the open-source project SpiceDB. This infrastructure supports high-performance demands of AI with billions of access control lists and millions of authorization checks per second, ensuring global replication across data centers. AuthZed's solutions are geared towards maintaining secure, scalable, and efficient AI operations.