Stock Markets April 28, 2026 12:26 PM

FDA Opens Pilot to Test Artificial Intelligence in Early Drug Development

Agency invites limited participants to demonstrate AI use cases for preclinical and Phase I decisions under a 30-day application window

By Leila Farooq
FDA Opens Pilot to Test Artificial Intelligence in Early Drug Development

The U.S. Food and Drug Administration has launched a pilot program to evaluate the use of artificial intelligence and data science in regulatory decision-making for drug and biologic development, focusing on preclinical and early clinical stages. The agency will publish the notice in the Federal Register on April 29 and is accepting applications through May 29. Selected companies will work with the FDA to assess model credibility, validation, and fitness for purpose to inform internal frameworks rather than to approve AI tools directly.

Key Points

  • FDA published notice in the Federal Register as FR Doc. 2026-08281 and opened a 30-day application window through May 29 for a pilot program on AI in drug development.
  • Selected companies will present AI use cases focused on preclinical and Phase I activities, including translational modeling, dose selection, early trial design, and biomarker strategies.
  • The FDA's objective is to build experience and internal frameworks for evaluating AI-generated evidence - not to approve AI tools - with guidance aligned to the NIST AI Risk Management Framework.

The Food and Drug Administration has announced a pilot aimed at testing how artificial intelligence and related data science techniques can be used in regulatory decision-making for drug and biologic development.

The agency said the notice will be formally published in the Federal Register on April 29 as FR Doc. 2026-08281. The program establishes a 30-day application period that runs through May 29, during which sponsors may seek selection to participate.

FDA officials plan to choose a limited number of companies to submit concrete AI use cases tied to preclinical work and early clinical development, with particular emphasis on Phase I applications. Examples of the intended use cases include translational modeling, dose selection, early trial design, and strategies for biomarkers that support the move from nonclinical results into first-in-human studies.

Participation will be interactive: the FDA will work directly with the selected sponsors to evaluate model credibility, validation approaches, and whether particular models are fit for the purpose intended. The agency emphasized that the exercise is designed to build practical experience and to create consistent internal frameworks for evaluating evidence generated by AI, not to constitute approval of AI tools themselves.

The pilot program will be guided by principles aligned with the National Institute of Standards and Technology AI Risk Management Framework. In outlining the program's aims, the FDA said it seeks to determine how AI and data science might improve trial efficiency, strengthen safety monitoring, support dose selection decisions, and enable more informed early go/no-go determinations.

FDA materials highlighted early-phase clinical trials as a frequent bottleneck in drug development, describing them as periods marked by high uncertainty, small and often limited patient populations, and decision-making processes that can be inefficient. The pilot is intended to address those specific challenges by testing whether AI-based approaches can provide useful, validated inputs in that context.

Elizabeth Anderson, an analyst at Evercore ISI, described the pilot as a "logical next step from the FDA" that begins to chart a practical pathway for AI's role in clinical development. Anderson said she expects AI adoption will proceed gradually, with near-term milestones driven by the pilot window and any subsequent guidance from the agency. She added that broader adoption across pharmaceutical and biotech companies is likely to unfold over two or more years and will depend on individual firms' decisions.


Context and next steps

Sponsors interested in participating must submit applications within the 30-day window ending May 29. Following selection, the FDA will engage with companies to examine specific AI use cases across preclinical and early clinical development to assess credibility and fitness for regulatory consideration.

Risks

  • The pilot addresses early-phase clinical trials, which the FDA noted are characterized by high uncertainty and limited patient populations - factors that can complicate model validation and regulatory assessment.
  • Selection of only a limited number of companies means insights will be based on a small set of use cases, potentially limiting the generalizability of findings in the near term.
  • The program is intended to inform frameworks rather than to approve AI tools, so adoption across pharma and biotech may be gradual and dependent on company-specific decisions over a multi-year horizon.

More from Stock Markets

OpenAI Expands Model Access to Amazon Bedrock, Adds Codex and Agent Tools on AWS Apr 28, 2026 Volkswagen in Advanced Talks to Repurpose Osnabrueck Plant Toward Defense Work Apr 28, 2026 Google Withdraws From $100 Million Pentagon Drone-Swarm Competition Apr 28, 2026 OpenAI and AWS Deepen Partnership to Deliver Models Inside Enterprise Cloud Environments Apr 28, 2026 Longi Green Energy’s Q1 Loss Widens Despite Rising Module Prices Apr 28, 2026