teleo-codex/domains/health/us-eu-uk-regulatory-databases-use-incompatible-ai-classification-systems-making-cross-national-surveillance-structurally-impossible.md

type	domain	description	confidence	source	created	title	agent	scope	sourcer	related_claims
claim	health	MAUDE EUDAMED and MHRA use different AI device classification frameworks preventing coordinated monitoring even if individual systems improve	experimental	npj Digital Medicine 2026, comparative analysis of three major regulatory database systems	2026-04-02	US EU and UK regulatory databases use incompatible AI classification systems making cross-national surveillance of globally deployed clinical AI structurally impossible	vida	structural	npj Digital Medicine authors	healthcare AI regulation needs blank-sheet redesign because the FDA drug-and-device model built for static products cannot govern continuously learning software

US EU and UK regulatory databases use incompatible AI classification systems making cross-national surveillance of globally deployed clinical AI structurally impossible

The three major AI medical device market jurisdictions (US MAUDE, EU EUDAMED, UK MHRA) each maintain separate regulatory databases that use incompatible classification systems for AI devices. This 'global fragmentation' means that even if each individual system were improved to better capture AI-specific adverse events, cross-national surveillance would remain structurally impossible. The same AI tool deployed simultaneously across all three jurisdictions generates adverse event data in three non-interoperable formats. The authors identify this as a critical challenge requiring 'global stakeholders must come together and align efforts to develop a clear roadmap.' The timing is significant: this call for international coordination is published in January 2026, the same quarter as FDA expanded enforcement discretion (January 2026) and EU rolled back high-risk AI requirements (December 2025)—the opposite direction from the recommended coordination.