Two-of-Three Security Rule

Definition

A heuristic for agentic AI safety: an agent may have any two of three capabilities (access sensitive information, execute code, communicate externally) at a time but never all three simultaneously.

Cast within

No topic-region cast yet — this appears once Two-of-Three Security Rule's compiled claims are aligned into a topic region's argument tree.

Claims mentioning Two-of-Three Security Rule (1)

Your notes

Loading notes...