Navigating the Future of GMP: Reliance, Annex 1, and AI
The pharmaceutical industry is at a crossroads, with evolving regulatory landscapes and emerging technologies like AI demanding a modernization of Good Manufacturing Practices (GMP). This article examines the impact of increased regulatory reliance, the implementation of EU GMP Annex 1, and the integration of Artificial Intelligence on future compliance strategies.
Executive Summary
- The EMA is actively pursuing regulatory reliance to reduce duplication without compromising quality standards.
- Annex 1 implementation requires a clear understanding of risk, scientific reasoning, and evidence of compliance, particularly for sterile products.
- AI integration in GMP necessitates strong governance, data control, traceability, and human oversight.
- EMA Annex 22 represents the first dedicated EU GMP guidance on AI, establishing compliance and validation expectations for static AI models in GMP systems.
- The EU AI Act adds a horizontal regulatory layer that intersects with GMP obligations for high-risk AI applications in pharmaceutical manufacturing.
Market Impact
| Regulatory | medium |
|---|---|
| Commercial | medium |
| Competitive | low |
| Investment | low |
Ask about this article
AI-assisted answers grounded in NovaPharmaNews intelligence
Answers use retrieved site intelligence plus AI synthesis. Verify critical decisions with primary sources.
Navigating the Future of GMP: Reliance, Annex 1, and AI
The pharmaceutical industry is at a crossroads, with evolving regulatory landscapes and emerging technologies like AI demanding a modernization of Good Manufacturing Practices (GMP). This article examines the impact of increased regulatory reliance, the implementation of EU GMP Annex 1, and the integration of Artificial Intelligence on future compliance strategies. With EMA Annex 22 GMP guidance now in development and the EU AI Act adding a horizontal layer of oversight, manufacturers face a compliance environment that is shifting faster than at any point in recent memory.
Key Takeaways
- The EMA is actively pursuing regulatory reliance to reduce duplication without compromising quality standards.
- Annex 1 implementation requires a clear understanding of risk, scientific reasoning, and evidence of compliance, particularly for sterile products.
- AI integration in GMP necessitates strong governance, data control, traceability, and human oversight.
- EMA Annex 22 represents the first dedicated EU GMP guidance on AI, establishing compliance and validation expectations for static AI models in GMP systems.
- The EU AI Act adds a horizontal regulatory layer that intersects with GMP obligations for high-risk AI applications in pharmaceutical manufacturing.
How Is the EMA Advancing Regulatory Reliance and Annex 1?
Brendan Cuddy, a lead scientific officer at the EMA and chair of the agency's Good Manufacturing and Distribution Practice Inspectors Working Group, has been central to efforts to expand regulatory reliance, support Annex 1 implementation, and develop the agency's thinking on AI in GMP environments. Speaking at the 2026 ISPE Europe Annual Conference, Cuddy outlined how the EMA and its international partners are working to reduce duplication without lowering standards — and where more legal, operational, and technical work is still needed.
Regulatory reliance, in the EMA's framework, means that one authority can place varying degrees of reliance on the assessments, inspections, and decisions made by another trusted authority. The goal is to avoid redundant inspections of the same manufacturing sites, accelerate medicine availability, and allocate scarce inspectorate resources more efficiently. Cuddy noted that several successful initiatives have already demonstrated the model's viability, though legal frameworks in some jurisdictions still limit how far reliance can be formalized.
On Annex 1, the revised EU GMP guidance for sterile medicinal products, Cuddy struck a pragmatic tone. The updated annex does not demand perfection overnight. But it does expect manufacturers to demonstrate a clear understanding of risk, solid scientific reasoning, and documented evidence of compliance. The guidance focuses on eliminating contamination risks during manufacturing — and critically, after the product leaves the cleanroom. For companies producing sterile injectables, ophthalmic products, and other aseptic preparations, the implications touch everything from facility design and environmental monitoring to personnel gowning and process validation.
Cuddy emphasized that inspector training is a key enabler for consistent Annex 1 enforcement across EU member states. Without harmonized interpretation, the risk of divergent expectations between national competent authorities increases — undermining the very consistency that reliance frameworks are designed to promote.
What Does Annex 22 Mean for AI in Pharmaceutical Manufacturing?
The EMA's development of Annex 22 — the first dedicated EU GMP guidance on artificial intelligence — signals that AI is no longer a theoretical consideration for pharmaceutical quality systems. A multistakeholder workshop convened by the EMA brought together industry, regulators, and technical experts to shape the annex's scope and requirements.
From the opening paragraphs of the draft, it is clear the annex applies primarily to static AI models deployed in GMP systems that could impact product quality, patient safety, or data integrity. This is a deliberate boundary. Static models — those that do not continuously learn or adapt in production — present a more tractable validation challenge than dynamic or continuously learning systems. The EMA's approach prioritizes compliance, validation, and risk control for safe, transparent AI deployment.
AI can operate in GMP environments through governed systems that use bounded data access, full traceability, and clear accountability. Generative and agentic models can support quality activities when prompts, data sources, logs, and human oversight are all defined and controlled. In pharmaceutical GMP settings, AI and machine learning can optimize batch production, enable predictive maintenance, improve process control, and facilitate real-time release testing — but only when the underlying data pipelines are validated and the model's decision boundaries are understood.
Research published in PMC underscores that AI/ML implementation in GMP requires a lifecycle approach spanning model development, training data governance, deployment, monitoring, and periodic revalidation. The EU AI Act adds another layer, classifying certain AI applications in regulated industries as high-risk and imposing additional transparency and conformity assessment obligations. Manufacturers should expect the intersection of GMP guidance and horizontal AI regulation to tighten over the next several years.
The EMA has signaled that Annex 22 will work in concert with the existing Annex 11 framework on computerized systems, creating a layered governance structure where general IT validation principles meet AI-specific requirements around model transparency, bias detection, and performance monitoring.
What Should Pharmaceutical Manufacturers Prioritize Now?
The practical impact of these three converging forces — reliance, Annex 1, and AI — varies by company profile, but several strategic imperatives apply broadly.
Embrace reliance as an operational advantage. Companies with manufacturing networks spanning multiple regions should actively engage with reliance frameworks. When a trusted authority inspects a site and that inspection is recognized by other regulators, the result is fewer duplicative audits, faster product approvals, and lower compliance overhead. Manufacturers that invest in inspection readiness and transparent quality management systems will be best positioned to benefit.
Treat Annex 1 compliance as a strategic investment, not a checkbox exercise. The revised annex demands a contamination control strategy that is scientifically justified and evidence-based. Companies that have deferred facility upgrades or lack rigorous environmental monitoring programs face significant catch-up costs. Those acting early — upgrading cleanroom classifications, implementing rapid microbial monitoring technologies, and training personnel to the new standard — will gain a competitive edge in regulatory inspections and customer audits.
Build AI governance before scaling AI deployment. The temptation to pilot AI tools across manufacturing and quality operations is strong, but Annex 22 and the EU AI Act both demand that governance precede scale. Manufacturers need clear policies on data integrity for AI training sets, model validation protocols, change control for model updates, and human-in-the-loop oversight requirements. Companies establishing these frameworks now will avoid costly rework when Annex 22 becomes enforceable.
The competitive landscape is shifting. Regulatory agencies are signaling that the future of GMP oversight will be more collaborative, more data-driven, and more technologically sophisticated. Manufacturers investing in cross-functional quality teams, digital infrastructure, and regulatory intelligence will navigate this transition more effectively than those waiting for enforcement actions to force their hand.
How Will the EU AI Act Reshape GMP Compliance Requirements?
The EU AI Act introduces a horizontal regulatory framework cutting across all regulated industries, including pharmaceuticals. For GMP purposes, the Act's classification of certain AI systems as "high-risk" triggers conformity assessment obligations that sit alongside existing EMA guidance. Manufacturers deploying AI for batch release decisions, predictive quality control, or automated deviation classification may find their systems subject to both Annex 22 and the AI Act.
This dual-regulatory reality demands that quality teams develop fluency in both frameworks. The EMA's GMP guidance hub provides the GMP-specific lens, while the AI Act establishes broader obligations around transparency, human oversight, and post-market monitoring. Companies mapping the overlap between these frameworks now — rather than treating them as separate compliance exercises — will reduce redundancy and build more defensible quality systems.
One area to watch is how the EMA coordinates with national competent authorities on AI Act enforcement within the pharmaceutical sector. If member states adopt divergent interpretations, manufacturers could face the same fragmentation that reliance frameworks were designed to prevent. Early engagement through industry working groups and public consultations remains the most effective path to coherent implementation.
Frequently Asked Questions
How is AI being integrated into GMP environments?
AI operates in GMP through governed systems with controlled data access, traceability, and accountability, supporting quality activities with defined prompts, data sources, logs, and human oversight.
What is the primary focus of EU GMP Annex 1?
EU GMP Annex 1 provides specific guidance for the manufacture of sterile medicinal products, aiming to eliminate contamination risks during and after manufacturing processes.
What are the key drivers for modernizing GMP oversight?
The increasing globalization, digitalization, and complexity of pharmaceutical manufacturing necessitate regulatory approaches that address international supply chains, advanced manufacturing science, and new technologies like AI.
What is EMA Annex 22?
Annex 22 is the EMA's first dedicated GMP guidance on artificial intelligence, establishing compliance, validation, and risk control requirements for static AI models used in GMP systems that could impact product quality or patient safety.
How does the EU AI Act affect pharmaceutical GMP?
The EU AI Act classifies certain AI applications in regulated industries as high-risk, imposing additional transparency, conformity assessment, and governance obligations that intersect with existing GMP requirements.
Related coverage
This article follows our editorial standards. Report a correction via editorial contact.
