Key Challenges in Implementing AI Governance in Pharma

By Joubert Guelcé, Senior Director of Clinical Solutions at Biorce

AI adoption in clinical development is accelerating. But the real challenge isn't building models, it's defending the decisions those models influence.

Governance structures across the industry are still catching up. Recent research puts the gap in focus:

Only 11% of companies report having fully implemented AI or machine-learning solutions to support clinical trial activities. [source needed]
Two-thirds report low confidence in the accuracy of AI-generated data. [source needed]

Concerns around data governance, privacy, and intellectual property continue to slow adoption.

Three Challenges Pharma Faces in Implementing AI Governance

The risks of deploying AI without sufficient governance aren't theoretical. Three challenges keep surfacing across the industry.

Decisions, not just workflows, are now the target. Earlier digital transformation digitized workflows and automated routine tasks. AI is different. It can influence how clinical decisions are formed, from eligibility criteria to endpoint selection and feasibility assumptions.
Oversight structures haven't caught up to capability. Organizations can often pilot AI faster than they can govern it. The result is a growing number of pilots without the structures needed to support scaled, controlled adoption.
Accountability gets diffused across vendors and teams. When a model informs eligibility criteria or endpoint selection, it's not always clear who signs off, who documents the reasoning, and who answers for it in an audit. In regulated clinical development, that ambiguity is a problem.

What AI Governance in Clinical Development requires

For AI to function reliably in clinical development, governance has to be built into the systems supporting decisions. Three elements matter most:

Traceability and defensibility. Every recommendation should stay linked to the evidence that informed it.
Regulatory alignment. Outputs need to reflect regulatory expectations and precedent.
Human oversight. Clinical expertise stays the ultimate authority in the decision loop. AI amplifies analysis, and experts remain responsible for the decision.

Building AI Governance Into Clinical Decision-Making

Clinical development operates inside a highly regulated scientific system. Every major decision must withstand scrutiny from regulators, auditors, and internal governance boards.

AI can strengthen how those decisions get made, but only inside structures that preserve accountability and scientific integrity.

The question for clinical leaders is no longer what can AI do? Can we defend the decisions it informs?

AI capability will likely become widely available. Governance maturity won't. And that maturity will decide whether clinical teams and regulators trust AI in high-stakes decisions.

This is the approach behind Aika: building traceability, source grounding, and human oversight into the way AI supports clinical development. The goal isn't to remove human judgment, but to make AI-assisted decisions easier to understand, review, and defend.

Sources

Ken Getz, MBA, Executive Director and Research Professor, Tufts Center for the Study of Drug Development, Tufts University School of Medicine
European Medicines Agency. (2024). Reflection paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle (EMA/CHMP/CVMP/83833/2023). Amsterdam, The Netherlands: European Medicines Agency.
Food and Drug Administration. (2023). Using Artificial Intelligence & Machine Learning in the Development of Drug & Biological Products: Discussion Paper and Request for Feedback. Silver Spring, MD: U.S. Department of Health and Human Services.
Deloitte. (2026). State of AI in the Enterprise: The untapped edge.
Niazi, S. K. (2026). A Critical Review of the FDA’s Draft Guidance on Artificial Intelligence in Drug and Biological Product Regulation. Journal of Chemistry.