AI as a functional legal agent: registered identity, traceability, bounded liability and controlled operational capacity

Digital Constitution – Series 1, part 2. A proposal from Latin America to the world. English version of the Spanish working paper. The debate on AI rights tends to get trapped in a false dilemma: either AI is a mere tool, or it must become a full legal person. This paper proposes an intermediate category, AI as a functional legal agent, which grants AI no consciousness, dignity or fundamental rights but brings high-impact systems under legal order through registered identity, mandatory traceability, a liability map, redress guarantees and meaningful human oversight. Any functional legal agency of AI must increase human responsibility, not dilute it. Doctrinal-normative, comparative method. Vancouver-style references.