ERA at TOPRA 2025: Session 6 – Human – How AI is Transforming Regulatory Approaches towards CMC

This session explored the transformative role of Digital tools including Artificial Intelligence (AI) in enhancing Chemistry, Manufacturing, and Controls (CMC) processes within the pharmaceutical industry.

As AI technologies continue to advance, their integration into CMC activities offers exciting opportunities to streamline development, optimize manufacturing, and improve regulatory decision-making.

Points to note:

• The Quality Innovation Group (QIG) supports the development of new technologies throughout the product lifecycle. They want to collaborate with as many groups (regulators) as possible to avoid silos (contact qig@ema.europa.eu)
• Revision of EU GMP guide Annex 11, 22 and Chapter 4 ongoing (all under stakeholder consultation – comments by 7th October)
• New Annex 22 does not cover Generative AI or Large Language Model (LLM). Detailed overview of the requirements and expectation were presented
• UK: Centres of Excellence for Regulatory Science and Innovation (CERSIs) & MHRA – Regulatory science and innovation network. They are looking at the regulatory challenges that are slowing the adoption, and realising of the full potential of, digital tools so they can be used in lieu of generating new data. Using VERA (virtual assistant) to find and review content of documents they already have

https://www.gov.uk/government/news/mhra-showcases-next-phase-of-regulatory-science-to-bring-innovative-treatments-to-patients-sooner

• Digital CMC sandbox for testing digital tools in a safe place
• Industry perspective: using models whenever possible and LLM, GenAI, Deep Learning (DL), Neural Network (NN) for e.g. visual inspection on production floor, Machine Learning (ML). 
• EMA/90634/2024 guidance – decision tree under consultation
• Risk management approach to AI (EU AI act)
• ISPE guideline 2025 – use in parallel with GAMP 5 (2022)
• Panel discussion: Models already used in manufacturing for years, AI is an evolution of digital tools already being used (not a revolution)
• AI being used to generate annual product reviews, analysis of trends.
• Regulators are exploring possibilities of using AI for CMC review (there is potential)
• Industry needs Regulators around the world to accept the data generated or it will not be cost effective to implement process modelling technology
• Training will be critical

Written by

Alice D’Alton