Life Sciences AI Is Moving at Two Different Speeds

Two clocks, one regulator

Life sciences AI is not moving at one speed. It is moving at two, and the gap between them is now the most useful signal for anyone setting AI governance budget for 2026.

On the device side, the clock is ticking fast and predictably. In the past two weeks alone, Caristo Diagnostics received FDA de novo authorization for an AI tool that quantifies coronary inflammation to help assess heart attack risk in patients without obstructive coronary artery disease MedTech Dive. DeepHealth secured 510(k) clearance for an AI platform that reads breast ultrasound images and generates BI-RADS-characterized reports MedTech Dive, a clearance detailed further by HIT Consultant. Vexev, meanwhile, closed $6 million to push its autonomous robotic ultrasound platform through remaining FDA regulatory steps after positive US clinical results BioSpace. These are not edge cases. They are the normal operation of an established Software as a Medical Device pathway that FDA and sponsors both understand.

On the drug discovery side, the clock has barely started. Despite roughly $8.9 billion in venture and pharma investment, AI-native drug discovery has zero FDA approvals to show for it, and the reason is not the biology Clinical Trial Vanguard. The platforms that generated the most impressive preclinical speed often did so by skipping the documentation architecture a submission requires, which means the speed advantage evaporates the moment a sponsor needs to defend model provenance, training data lineage, or performance claims to a reviewer.

Regulators are signaling where that documentation bar will land. In January 2026, FDA and EMA jointly published ten principles for good AI practice in drug development, covering data governance, model performance, and human oversight Forbes. FDA’s draft guidance on AI-generated evidence outlines how such evidence can support regulatory decisions, a framework one industry expert calls the key reference point for anyone building trial design around AI Technology Networks. A 2025 FDA framework and an EMA reflection paper both point toward eventual acceptance of simulation and synthetic control arms, without yet opening the door fully Calcalist.

What this means for governance spend

The device pathway rewards companies that already know how to build a 510(k) or de novo file. The discovery pathway rewards companies that build submission-grade documentation now, before there is a mandate to do so. Waiting for FDA and EMA to finalize the ten principles into binding guidance means competing for review bandwidth against every other sponsor who waited too. The bill referenced by Clinical Trial Vanguard comes due for whoever has the weakest audit trail when the door does open, not whoever has the fastest model.

For compliance leaders managing both device and discovery portfolios, the near-term move is straightforward. Fund SaMD documentation at the pace regulatory precedent already supports. Fund AI-native discovery documentation ahead of precedent, because that clock is about to start running whether the architecture is ready or not.

---

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 thesis—device AI moves fast on established pathways while drug discovery AI lacks approvals despite investment—is coherent and supported, but the causal claim that discovery’s lag stems prima
Source & Claim Verification	Qwen · local	cleared. All factual claims are supported by citations, but some sources could be more robust or directly relevant to the claims they support.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately reflects FDA SaMD pathways and emerging AI drug discovery expectations, but omits explicit alignment with ISO 42001, EU AI Act, and MDR/IVDR requirements.
Technical Accuracy	Llama	cleared. The article is generally accurate in its representation of the current state of Life Sciences AI, but contains some minor inaccuracies and dated references due to the fictional future date.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively distinguishes between vendor hype and regulatory reality, particularly in the drug discovery section, by highlighting the lack of FDA approvals despite significant investment.
Novelty & Non-Duplication	Grok	held. The two-clocks governance frame is a light synthesis, but the substance largely restates wire clearances plus the already-circulating $8.9B/zero-approvals critique rather than adding non-duplicative i
Validation	DeepSeek	cleared. The central claim that AI in life sciences is moving at two distinct speeds—fast for devices and slow for drug discovery—is strongly validated by recent, specific FDA authorizations for devices versus

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