Regulators set AI rules
What happened
- FDA and the European Medicines Agency issued joint ten principles to guide AI use across the medicines lifecycle. - The FDA also deployed an agency-wide agentic AI platform used for reviews, surveillance and inspections. - Regulators operationalised agentic AI while emphasising strict data protections, human oversight, and lifecycle monitoring. (appliedclinicaltrialsonline.com) (appliedclinicaltrialsonline.com)
Why it matters
The U.S. Food and Drug Administration and the European Medicines Agency have moved from talking about artificial intelligence in drug regulation to setting joint rules and using the technology inside their own agencies. (ema.europa.eu) On January 14, 2026, the two regulators published 10 joint principles for “good AI practice” across the medicines lifecycle, from early research and clinical trials to manufacturing and post-market safety monitoring. The principles apply to drugmakers, applicants and marketing authorization holders using AI to generate evidence for regulators. (fda.gov) The principles call for a risk-based approach, data quality controls, model transparency, human oversight, privacy and cybersecurity protections, and monitoring after deployment so systems are checked as conditions change. The agencies said the framework is meant to guide evidence generation rather than serve as a detailed binding rulebook. (ema.europa.eu) In plain terms, regulators are treating AI models like other tools used to support drug decisions: developers must show the data are reliable, the model fits its purpose, and people remain accountable for the outcome. That approach follows a September 2024 European Medicines Agency reflection paper that laid out how AI could be used in development, authorization and post-authorization work. (ema.europa.eu) The U.S. Food and Drug Administration is also using AI internally. On December 1, 2025, it said it had deployed agentic AI capabilities for all agency employees, after earlier launching Elsa, a generative AI tool for staff including scientific reviewers and investigators. (fda.gov) The agency said those tools are being used in scientific reviews, surveillance and inspections, and that the systems run inside a secure government environment rather than on public consumer chatbots. The Food and Drug Administration said the rollout followed a pilot of AI-assisted scientific review and is being coordinated by Chief AI Officer Jeremy Walsh and Sridhar Mantha. (fda.gov) This shift comes as the Food and Drug Administration says drug submissions with AI components have increased in recent years across nonclinical, clinical, manufacturing and postmarketing phases. The agency has built a dedicated artificial intelligence for drug development program inside the Center for Drug Evaluation and Research to track that change. (fda.gov) The Food and Drug Administration has also been filling in the policy details. A draft guidance on AI used to support regulatory decision-making for drugs and biologics recommends sponsors define the model’s context of use, assess credibility and manage risks from data drift and performance changes over time. (fda.gov) The European Medicines Agency has taken a similar line on large language models and other AI systems: useful for efficiency, but only with documented controls, secure handling of data and review by trained staff. The joint U.S.-European principles now give drug developers a common baseline on both sides of the Atlantic. (ema.europa.eu) What happens next is less about a single new law than about how these principles get translated into reviews, inspections and submissions. The message from both agencies is that AI can speed work on medicines, but the humans signing the decision still own it. (fda.gov)
Key numbers
- (ema.europa.eu) On January 14, 2026, the two regulators published 10 joint principles for “good AI practice” across the medicines lifecycle, from early research and clinical trials to manufacturing and post-market safety monitoring.
- That approach follows a September 2024 European Medicines Agency reflection paper that laid out how AI could be used in development, authorization and post-authorization work.
- On December 1, 2025, it said it had deployed agentic AI capabilities for all agency employees, after earlier launching Elsa, a generative AI tool for staff including scientific reviewers and investigators.
What happens next
- That approach follows a September 2024 European Medicines Agency reflection paper that laid out how AI could be used in development, authorization and post-authorization work.
- (ema.europa.eu) What happens next is less about a single new law than about how these principles get translated into reviews, inspections and submissions.
Quick answers
What happened in Regulators set AI rules?
FDA and the European Medicines Agency issued joint ten principles to guide AI use across the medicines lifecycle. The FDA also deployed an agency-wide agentic AI platform used for reviews, surveillance and inspections. Regulators operationalised agentic AI while emphasising strict data protections, human oversight, and lifecycle monitoring. (appliedclinicaltrialsonline.com) (appliedclinicaltrialsonline.com)
Why does Regulators set AI rules matter?
The U.S. Food and Drug Administration and the European Medicines Agency have moved from talking about artificial intelligence in drug regulation to setting joint rules and using the technology inside their own agencies. (ema.europa.eu) On January 14, 2026, the two regulators published 10 joint principles for “good AI practice” across the medicines lifecycle, from early research and clinical trials to manufacturing and post-market safety monitoring. The principles apply to drugmakers, applicants and marketing authorization holders using AI to generate evidence for regulators. (fda.gov) The principles call for a risk-based approach, data quality controls, model transparency, human oversight, privacy and cybersecurity protections, and monitoring after deployment so systems are checked as conditions change. The agencies said the framework is meant to guide evidence generation rather than serve as a detailed binding rulebook. (ema.europa.eu) In plain terms, regulators are treating AI models like other tools used to support drug decisions: developers must show the data are reliable, the model fits its purpose, and people remain accountable for the outcome. That approach follows a September 2024 European Medicines Agency reflection paper that laid out how AI could be used in development, authorization and post-authorization work. (ema.europa.eu) The U.S. Food and Drug Administration is also using AI internally. On December 1, 2025, it said it had deployed agentic AI capabilities for all agency employees, after earlier launching Elsa, a generative AI tool for staff including scientific reviewers and investigators. (fda.gov) The agency said those tools are being used in scientific reviews, surveillance and inspections, and that the systems run inside a secure government environment rather than on public consumer chatbots. The Food and Drug Administration said the rollout followed a pilot of AI-assisted scientific review and is being coordinated by Chief AI Officer Jeremy Walsh and Sridhar Mantha. (fda.gov) This shift comes as the Food and Drug Administration says drug submissions with AI components have increased in recent years across nonclinical, clinical, manufacturing and postmarketing phases. The agency has built a dedicated artificial intelligence for drug development program inside the Center for Drug Evaluation and Research to track that change. (fda.gov) The Food and Drug Administration has also been filling in the policy details. A draft guidance on AI used to support regulatory decision-making for drugs and biologics recommends sponsors define the model’s context of use, assess credibility and manage risks from data drift and performance changes over time. (fda.gov) The European Medicines Agency has taken a similar line on large language models and other AI systems: useful for efficiency, but only with documented controls, secure handling of data and review by trained staff. The joint U.S.-European principles now give drug developers a common baseline on both sides of the Atlantic. (ema.europa.eu) What happens next is less about a single new law than about how these principles get translated into reviews, inspections and submissions. The message from both agencies is that AI can speed work on medicines, but the humans signing the decision still own it. (fda.gov)
