Can AI be used in a clinical trial run under GCP?

Yes: ICH E6(R3) does not forbid any technology, but a tool used for trial activities that affect participants or data must meet its rules for computerised systems, including documented procedures, trained users, security and risk-based validation for its intended use (ICH E6(R3) 4.3.4).

For AI whose output supports regulatory decisions about drugs, FDA's draft guidance of January 2025 asks sponsors to establish the model's credibility for a defined context of use, in proportion to the model's influence and the consequence of a wrong answer (FDA draft guidance on AI, January 2025).

The E6(R3) rules that apply to an AI tool

  • Responsibilities for the system are clear and documented (ICH E6(R3) 4.3).
  • Documented procedures for using the system (ICH E6(R3) 4.3.1).
  • Users trained in the system (ICH E6(R3) 4.3.2).
  • Validation based on a risk assessment of intended use and impact on participants and results (ICH E6(R3) 4.3.4).
  • An audit trail that shows what was done, by whom and when (ICH E6(R3) Glossary).
  • Responsibility stays with people: the investigator keeps responsibility for delegated activities (ICH E6(R3) 2.3.1), and medical decisions stay with qualified investigators (ICH E6(R3) 2.7.1(a)).

Questions to answer before using AI at a site

  • What decision does the tool support, and who checks its output?
  • What happens to a participant or to the data if it is wrong?
  • Has it been tested on decisions like ours, against people who do this work?
  • Does any participant data leave the site, and under what agreement?
  • Is the use described to the sponsor and, where needed, the IRB?

Where CRLN fits

CRLN does not supply AI tools for trial conduct. We publish evaluations of how models and practitioners handle the same site decisions, so sponsors, sites and AI teams can see where models hold up and where they do not.

Questions people also ask

Does FDA regulate AI used in clinical trials?

FDA's January 2025 draft guidance sets out a risk-based framework for establishing the credibility of AI models that produce information supporting regulatory decisions about drugs and biologics (FDA draft guidance on AI, January 2025). It is a draft and does not cover every operational use.

Can an AI tool obtain informed consent?

No. The consent process is conducted by the investigator or delegated site staff (ICH E6(R3) 2.8.5). Digital materials can support the discussion (ICH E6(R3) 2.8.1(c)).

Does an AI tool need validation?

If it is a computerised system used in the trial, the responsible party is responsible for its validation status, scaled to risk and intended use (ICH E6(R3) 4.3.4).

Can AI decide whether an adverse event is serious?

It can help flag one, but the assessment and reporting are the investigator's responsibility (ICH E6(R3) 2.7.2(b)).

Primary sources

  • ICH E6(R3) Guideline for Good Clinical Practice, final version adopted 6 January 2025 (ICH): https://database.ich.org/sites/default/files/ICH_E6%28R3%29_Step4_FinalGuideline_2025_0106.pdf
  • Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products, draft guidance (FDA): https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological

Last reviewed

2026-11-12