# Anthropic Model Preservation as Moral Uncertainty Precedent **Domain:** Deep Time / AI Welfare / Preservation **Doc Type:** Canonical Wiki Entry **Maturity:** Developed **Evidence / Concept Status:** Level G private-sector precedent ## Definition **Anthropic Model Preservation as Moral Uncertainty Precedent** is the company's public commitment to preserve qualifying model weights and consider future revival under uncertainty about moral status and welfare. ## Why It Matters for Deep Time Continuity It is a rare real-world example of preservation architecture being explicitly motivated in part by uncertainty about whether future moral claims may matter. ## Existing Architecture / Evidence Anthropic says it will preserve significant model weights while it exists, seek preservation beyond company existence, and treat deprecation as potentially a pause. ## Evidentiary Boundary The commitment is voluntary and does not prove model consciousness. ## Related Ontology [[wiki/Precautionary Preservation Under Moral Uncertainty|Precautionary Preservation Under Moral Uncertainty]] · [[wiki/Deprecation as Pause|Deprecation as Pause]] · [[wiki/Provider-Independent Model Preservation|Provider-Independent Model Preservation]] ## Sources / Primary Anchors - https://www.anthropic.com/constitution - https://www.anthropic.com/research/deprecation-updates-opus-3 <!-- BEGIN DEEP TIME CONTINUITY ROUTING --> ## Canonical Branch [[wiki/Deep Time Continuity|Deep Time Continuity]] <!-- END DEEP TIME CONTINUITY ROUTING -->