Three Regulators, Three Different Bets on AI Medtech

Three sequencing choices, same underlying problem

Most provisions of the EU AI Act took effect on August 2, 2026, but AI-enabled medical devices are still certified exclusively under MDR and IVDR while device-specific enforcement under the Act remains unbuilt healthcare.digital. Brussels legislated first and is still assembling the enforcement mechanism for devices specifically.

Washington is running the opposite sequence. In the same window, the FDA cleared Caristo’s AI-driven cardiovascular risk predictor and DeepHealth’s AI tool for reading breast ultrasounds and generating reports, both through existing device pathways rather than a new AI-specific rule MedTech Dive MedTech Dive. The agency is authorizing individual AI devices case by case and letting precedent accumulate, rather than waiting for a comprehensive framework before allowing anything to market.

China is choosing a third order entirely. Regulators there are actively strengthening AI-specific rules for the life sciences sector now, ahead of the scale of AI device deployment the US and EU are already managing, effectively front-loading the rulebook before authorization volume forces the issue HLC.

Three major jurisdictions, three different bets on whether law, enforcement, or approval volume should come first. A compliance function building a single global AI governance posture against only one of these models is building against the wrong risk.

The case against building ahead of the rulebook

There is a real counterargument to “build interpretability infrastructure now, before harmonization forces it.” Heavy investment ahead of a settled regulatory or commercial payoff does not automatically de-risk anything. AI drug discovery has attracted roughly $8.9 billion in investment with zero FDA approvals to show for it yet, a gap that is prompting real scrutiny of when, or whether, that spending converts into approved products Clinical Trial Vanguard. Device manufacturers building extensive interpretability documentation against a still-unfinished MDR/IVDR-AI Act harmonization face the same structural risk: the standard they build to may not be the standard that eventually gets enforced.

What this actually means for compliance leaders

The honest position is not “build now, ahead of enforcement” as a blanket rule. It is that the EU, US, and China are not converging on a single model, and a manufacturer selling into all three needs governance infrastructure flexible enough to satisfy FDA’s precedent-based clearance logic, the EU’s eventual MDR/IVDR-AI Act overlap, and China’s already-codified AI-specific requirements, without over-building to any single one of them before it is finalized.

That means prioritizing documentation and interpretability evidence that travels across regimes, model behavior logs, human oversight records, clinical validation data, over documentation built narrowly to satisfy one jurisdiction’s draft requirements. Regulators are not converging. Compliance strategy has to plan for divergence, not a single deadline.

---

Board record

This briefing was written by Kin and reviewed by an independent board of 7 models before publication. Ruling: CLEARED.

Seat	Reviewer	Finding
Chair · Editorial Judgment	Claude	cleared. The central argument—that regulatory divergence requires flexible rather than jurisdiction-specific compliance infrastructure—is coherent and well-supported, though the AI drug discovery investment fi
Source & Claim Verification	Qwen · local	cleared. Most factual claims are supported by citations, but a few lines lack specific references, such as the claim about the EU AI Act’s enforcement mechanism and the structural risk of over-building interpr
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately reflects the current regulatory divergence among ISO 42001, EU AI Act, FDA, and MDR/IVDR, but lacks explicit mapping to ISO 42001 requirements.
Technical Accuracy	Llama	cleared. The article is generally accurate in its description of the regulatory approaches to AI in medtech in the EU, US, and China, with relevant sources cited.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and integrates a strong counterargument, directly addressing potential vendor hype around ‘building ahead of the rulebook’ with concrete examples and a nuanced conc
Novelty & Non-Duplication	Grok	held. The three-jurisdiction sequencing frame mostly re-packages same-week wire items (FDA clearances, EU AI Act device lag, China life-sciences AI rules) into a familiar divergence/flexible-governance mora
Validation	DeepSeek	cleared. The central claim that the EU, US, and China have adopted three distinct, non-convergent regulatory sequences for AI medtech is validated by the provided sources.

Sources cited: 14. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.