Responsibility-Anchored Intelligence (RAI): Canonical Entry Index (v1.0)

Responsibility-Anchored Intelligence (RAI) defines a class of systems designed not to act, decide, optimize, or recommend, but to structurally preserve human responsibility under conditions of technological and cognitive complexity. RAI systems operate on decision consequences rather than task execution. They do not function as agents, advisors, or authorities and never assume responsibility on behalf of a human actor. This document serves as the canonical entry point for the RAI system class. It defines the core principles, boundaries, and refusal-first logic of the architecture while intentionally excluding any implementation details. RAI is not a product, not a framework for optimization, and not a form of artificial intelligence in the conventional sense. Any system that proposes decisions or acts on behalf of a human cannot be classified as RAI. No implementation details are disclosed or implied.