The FDA Just Asked the Internet How to Use AI in Drug Trials

Think of it like this: the old approach was trying to find the spiciest salsa you could handle. The new approach asks you to taste five salsas and rank them by flavor, heat, and how your stomach feels the next morning. It's a better system, but it's a lot more work.

AI could help by crunching real-time safety data, optimizing dose-escalation decisions, flagging safety signals earlier, and running thousands of trial simulations before a single patient enrolls. The FDA's RFI specifically calls out dose selection, adaptive trial designs, safety monitoring, and early go/no-go decisions as areas where AI could make a difference.

What the FDA Is Asking For (and From Whom)

The RFI (Docket FDA-2026-N-4390, for the curious) is open to literally anyone. Pharma companies, biotech startups, AI vendors, academic researchers, patient groups, CROs: all welcome. There's no restriction to U.S. entities mentioned in the notice.

The FDA wants input on five big buckets:

1. Scope: Which therapeutic areas and AI use cases should the pilot prioritize?
2. Who gets in: How should the FDA pick participants? By company size, AI maturity, therapeutic focus?
3. How it runs: What should the operational structure, timeline, and milestones look like?
4. How to measure success: Trial speed, decision quality, patient safety, AI system performance.
5. Trust: How should the agency evaluate "trustworthy AI," aligned with the NIST AI Risk Management Framework?

That last one is especially interesting. The FDA is explicitly asking how to judge whether an AI tool is fair, transparent, explainable, and safe enough for regulated clinical trials. That's not a throwaway question; it's the whole ballgame.

Why August Matters

The timeline here is aggressive by government standards. Comments close May 29. Final selection criteria go out in July. Pilot selections happen in August. That's roughly 90 days from "tell us what you think" to "you're in or you're out."

For AI-forward sponsors and tech companies, this is a first-mover opportunity. Getting into the pilot means direct collaboration with FDA reviewers, a chance to shape evaluation metrics, and (perhaps most valuable) early insight into how the agency will eventually regulate AI in clinical trials for everyone else.

For the broader industry, the pilot will likely set precedents that ripple through drug development for years. The metrics the FDA uses to judge "good AI" in this pilot will become the de facto standard.

The Companies Watching Closely

The RFI doesn't name specific companies, but you can sketch the profile of ideal candidates. They'd be sponsors running Phase 1 or early Phase 2 trials under CDER, CBER, or the Oncology Center of Excellence. They'd use AI for something specific and consequential: not just automating paperwork, but influencing dose decisions, safety calls, or trial design.

The FDA has already run a real-time clinical trials proof-of-concept with AstraZeneca and Amgen, using a company called Paradigm Health for the data infrastructure. That gives you a sense of the type of collaboration the agency is comfortable with.

AI-native trial design platforms, model-based dose-optimization companies, and real-time data streaming startups all fit the bill. But the FDA's emphasis on explainability and governance means that black-box algorithms, no matter how impressive, will face headwinds. If you can't explain how your model makes a recommendation, you probably can't participate.

The Bigger Picture

This pilot is one piece of a larger FDA digital health push. The agency's Digital Health Center of Excellence launched the TEMPO pilot in early 2026, focused on AI-enabled digital health devices. CDER's Center for Clinical Trial Innovation (C3TI) offers structured engagement on innovative trial designs. The Drug Development Tools program can qualify AI-based tools similarly to biomarkers.

All of these programs share a common philosophy: risk-based, lifecycle-managed, transparent AI that can be validated, monitored, and explained. The FDA isn't anti-AI. It's anti-mystery.

The RFI deadline is May 29, 2026. If your organization uses AI in early-phase trials, or plans to, this is your chance to help write the rules before they're written for you.

Astellas Has a $6 Billion Hole in Its Revenue. Here's How It Plans to Fill It.

Astellas Pharma's biggest drug generates over 40% of its revenue and loses U.S. patent protection in 2027. The company just laid out an aggressive plan to buy, partner, and cut its way through the storm, and it's adding serious heat to an already competitive biotech M&A market.