The Contamination Problem - Why AI Outputs Are Legally Unstable and How the Trace Economy Fixes It.

This whitepaper addresses the contamination problem in AI outputs — the systemic entanglement of generative models with proprietary, copyrighted, or restricted material. While AI is accelerating discovery in fields ranging from nanotechnology to pharmaceuticals, its outputs remain epistemically and legally unstable because lineage cannot be verified. Current responses — post-hoc source linking, copyright doctrines, patent standards, and licensing agreements — fail to resolve this liability. The paper introduces the Trace Economy and Proof of Cognitive Work (PoCW) protocol as the missing compliance layer. By anchoring outputs in timestamped human initiation, logging symbolic contributions as trace nodes, and redistributing royalties across the full lineage, the protocol converts contamination from a liability into a system of enforceable attribution. A case study of AI-designed carbon nanolattices illustrates how even apparently clean discoveries remain entangled with prior symbolic scaffolds, and how PoCW resolves these entanglements through trace logging. The paper concludes that without trace, AI outputs are contaminated by default; with trace, they become compliance-ready, royalty-redistributable, and future-proof.