Capital Is Flowing to AI. Regulatory Science Capacity Is Not.

The pattern behind three unrelated stories

A Nature analysis finds the EU’s early-stage medical device clinical study requirements are disharmonized across member states, and that this fragmentation is delaying patient access to innovation, not accelerating it Nature. A companion Nature piece proposes a fix: dedicated Centres of Excellence in Regulatory Science and Innovation that would generate early evidence and do horizon scanning before rules are locked in, rather than reacting after the fact Nature. Separately, commentary on Nature Medicine’s 2026 medical AI superintelligence framework argues FDA benchmarking guidance has not caught up to what current model capability claims actually require to evaluate Clinical Trial Vanguard.

The connective tissue is not that all three describe AI. Two don’t. It’s that all three describe regulators trying to evaluate fast-moving evidence with institutions built for a slower cadence, whether the object under review is a device trial protocol or a model’s benchmark claims. Fragmented device rules and outdated benchmarking guidance are the same failure mode wearing different clothes: the evidence-generation infrastructure hasn’t scaled with what it’s being asked to assess.

That gap has a live cost right now. Pharma companies are running AI-driven pipelines faster than they, or their regulators, can characterize the risk, according to recent industry reporting on deployment practices outpacing risk understanding HIT Consultant. Concrete cases back this up: Dasher Neuroscience has completed Phase 2 enrollment for an AI-derived drug candidate BioSpace, Insilico has moved an AI-designed mesothelioma compound forward Mesothelioma Guide, and Elix has partnered with the University of Vienna on AI-driven discovery GEN. None of this is illegitimate activity. It is evidence that the pipeline is real and the regulatory science underneath it is the part still catching up.

The counterargument, and where it falls short

The obvious rebuttal is that capital, not regulatory science, is Europe’s actual constraint. The EU has committed $11.4 billion to seven AI gigafactories specifically to close a compute gap with the US and China AP News. That is a real and defensible priority. But compute capacity and evaluation capacity are different bottlenecks, and only one of them has a Nature-documented backlog of disharmonized rules and no dedicated evidence-generation institution. Building gigafactories without a parallel investment in regulatory science capacity risks producing more AI-derived clinical evidence than any single member state’s device pathway, or the FDA’s current benchmarking framework, can actually adjudicate.

Note also that several sources here are vendor press releases, useful as evidence of pipeline momentum but not independent proof of regulatory readiness BioSpace, Mesothelioma Guide. Patients uploading their own data into AI tools are already navigating this gap without waiting for regulators to close it CNN.

Compliance leaders should treat regulatory science capacity as a line item, not an assumption. Ask whether your framework of record, MDR/IVDR, FDA guidance, or an internal ISO 42001 program, has kept pace with what you’re actually deploying.

---

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 core argument—that regulatory science capacity is a distinct bottleneck from capital/compute—is coherent and well-supported, though the claim that fragmented device rules and outdated benchmarking
Source & Claim Verification	Qwen · local	cleared. All factual claims are traced to citations, but some sources are vendor press releases which may not be fully independent.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately identifies regulatory gaps in ISO 42001, EU AI Act, and FDA/MDR/IVDR contexts but lacks specific citations or direct alignment with their technical requirements.
Technical Accuracy	Llama	cleared. The article accurately highlights the gap between the rapid advancement of AI in healthcare and the slower pace of regulatory science capacity, citing relevant sources and technical concerns.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and addresses a relevant counterargument, and explicitly calls out vendor-hype in its sourcing.
Novelty & Non-Duplication	Grok	held. The capital-vs-regulatory-science-capacity contrast is a serviceable synthesis that joins otherwise siloed Nature device-rule pieces with AI benchmarking and gigafactory spend, but the core “instituti
Validation	DeepSeek	cleared. The briefing’s central claim that regulatory science capacity is not scaling with AI-driven evidence generation is supported by multiple independent sources documenting institutional lag and fragmenta

Sources cited: 11. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.