On 29 October 2026, all 53 Member States of the WHO European Region unanimously adopted the Region’s first Roadmap for action on artificial intelligence in health. Covering 2026–2033, the roadmap establishes a common regional approach centred on governance, lifecycle oversight, data quality, interoperability, cybersecurity, workforce capability and real-world evaluation.
The roadmap is not pharmaceutical GMP guidance, but it is important for pharmaceutical companies because AI used in clinical development, digital health, real-world evidence, patient-support technologies and post-market activities increasingly operates inside the same health-system infrastructure that the roadmap is intended to govern.
A Common AI Framework for 53 Countries
The roadmap was adopted during the 76th session of the WHO Regional Committee for Europe.
It covers countries across Europe and Central Asia and represents the first common WHO/Europe regional strategy dedicated specifically to AI in health.
The objective is to support AI that improves:
- quality of care;
- health-system resilience;
- patient outcomes;
- health-workforce effectiveness;
- equitable access.
The roadmap covers the period:
2026–2033.
Four Priority Areas Structure the Roadmap
WHO/Europe identifies four connected priority areas:
- governance and leadership;
- ethical, regulatory and legal frameworks;
- data and digital infrastructure;
- capacity building, implementation support and learning.
This is important because AI adoption is treated as a system-level transformation rather than a technology-purchasing exercise.
Lifecycle Governance Is Explicitly Required
One of the roadmap’s central principles is lifecycle governance.
AI systems need controls extending beyond development and initial deployment.
A lifecycle approach may include:
Design → Development → Evaluation → Deployment → Monitoring → Change → Reassessment
This is highly consistent with emerging pharmaceutical AI principles.
The underlying concept is that an AI system cannot be considered permanently trustworthy because it performed adequately during initial testing.
AI Must Be Proven Safe and Effective Before Scaling
WHO/Europe stresses that AI tools should demonstrate appropriate safety and effectiveness before large-scale deployment.
This is an important distinction between:
technical feasibility
and
evidence supporting real-world use.
For pharmaceutical companies, a similar principle applies when AI influences:
- clinical research;
- patient selection;
- digital endpoints;
- safety monitoring;
- patient-support applications;
- real-world evidence generation.
A successful pilot is not automatically sufficient evidence for broad regulated use.
Data Quality Is a Foundation, Not an Afterthought
The roadmap identifies data quality as a foundation for trustworthy AI.
This includes the broader infrastructure required to ensure that data are:
- appropriate;
- accessible under controlled conditions;
- interoperable;
- secure;
- suitable for the intended AI application.
This reinforces an increasingly common regulatory principle:
AI quality cannot be separated from data quality.
Interoperability Becomes an AI Governance Issue
Health data are distributed across:
- electronic health records;
- clinical databases;
- registries;
- laboratory systems;
- public-health systems;
- pharmaceutical systems.
AI increasingly needs to work across these environments.
Poor interoperability can introduce:
- missing context;
- inconsistent terminology;
- incorrect mappings;
- duplicate records;
- misinterpreted clinical information.
Interoperability is therefore not only an IT issue.
It directly affects AI reliability.
Cybersecurity Is Part of Trustworthy AI
WHO/Europe places cybersecurity alongside data governance and digital infrastructure.
This is particularly relevant because AI systems may:
- process sensitive health information;
- connect to multiple external systems;
- use cloud infrastructure;
- depend on third-party APIs;
- support consequential health decisions.
An accurate AI model deployed in an insecure architecture cannot be considered trustworthy.
Human Accountability Remains Central
WHO/Europe emphasises that health care remains fundamentally human-centred.
AI should support health professionals and patients rather than eliminate responsibility.
This is consistent with a broader principle:
AI capability does not eliminate human accountability.
For pharmaceutical applications, the accountable person must still understand:
- what the AI contributes;
- what its limitations are;
- when its output should be challenged;
- who owns the final decision.
AI Literacy Is Identified as a Major Readiness Gap
WHO/Europe’s underlying readiness assessment found substantial AI adoption but much weaker workforce preparation.
According to WHO/Europe:
- nearly two-thirds of responding countries already use AI for diagnosis;
- about half use conversational AI;
- only around one in five teach AI to health students;
- only around one in four provide AI training to the existing workforce;
- fewer than half had assessed whether existing laws adequately cover AI;
- almost 40% lacked ethical guidance;
- only 8% had a health-specific AI strategy.
These figures illustrate an important implementation gap:
technology adoption is advancing faster than governance and competence.
This Has a Parallel in Pharmaceutical GMP
Pharmaceutical companies face a similar challenge.
AI can be made technically available very quickly.
But responsible use requires:
- training;
- role definitions;
- approved use cases;
- critical review skills;
- escalation routes;
- understanding of model limitations.
Giving employees access to AI without building AI competence can create a governance gap even when the underlying technology performs well.
The Roadmap Aligns with the EU AI Act and Other European Frameworks
WHO/Europe specifically aims to avoid regulatory fragmentation.
The roadmap aligns with instruments including:
- the EU Artificial Intelligence Act;
- the European Health Data Space;
- the Medical Device Regulation;
- the General Data Protection Regulation;
- European AI policy initiatives;
- OECD work on scaling AI in health.
This alignment matters to pharmaceutical companies operating across multiple European jurisdictions.
Duplicative or inconsistent AI frameworks could otherwise create substantial regulatory complexity.
Why Pharmaceutical Companies Should Monitor Health-System AI Governance
Many pharmaceutical AI applications do not remain entirely inside the company.
Examples include:
- digital biomarkers;
- patient-support applications;
- decentralised clinical-trial technologies;
- real-world-data platforms;
- clinical decision-support tools;
- AI-assisted pharmacovigilance;
- connected drug-delivery systems.
These systems interact with health-care organisations and national data infrastructures.
The regulatory environment surrounding health-system AI can therefore affect pharmaceutical deployment strategies.
Real-World Evidence May Be Particularly Affected
Pharmaceutical companies increasingly use health-system data for:
- post-authorisation studies;
- safety monitoring;
- external controls;
- comparative effectiveness;
- regulatory evidence generation.
If national systems improve:
- data quality;
- interoperability;
- governance;
- secure access,
the reliability of AI-assisted RWE may also improve.
Conversely, weak upstream health-data governance can limit what an AI model can credibly infer.
A Practical Assurance Model Emerges
The WHO roadmap can be translated into a practical sequence:
Governance
↓
Legal and Ethical Assessment
↓
Data Quality and Infrastructure
↓
AI Development and Evaluation
↓
Controlled Deployment
↓
Workforce Competence
↓
Real-World Monitoring
This is very different from a technology-first approach.
The Roadmap Comes with an Implementation Support Guide
WHO/Europe has also published an implementation support guide intended to help countries apply the roadmap at different levels of AI and digital maturity.
This is important because countries in the WHO European Region do not begin from the same technological or regulatory baseline.
The strategy therefore allows implementation to be scaled to local readiness while maintaining common principles.
A New WHO AI Capability Is Also Planned
WHO/Europe Member States supported establishment of a dedicated WHO office or capability focused on AI.
Its intended functions include:
- evidence generation;
- technical assistance;
- capacity building;
- knowledge exchange;
- implementation support.
This suggests that WHO intends the roadmap to become an active implementation programme rather than remain only a strategic document.
Implications for Pharmaceutical AI Strategy
Pharmaceutical companies should consider health-system AI governance when planning applications involving external clinical or patient data.
Relevant questions include:
- Which national AI governance rules apply?
- Are health data interoperable and sufficiently controlled?
- Can data provenance be reconstructed?
- Are users appropriately trained?
- What real-world monitoring is required?
- How does the application interact with the EU AI Act, GDPR and sector-specific regulation?
Why This Roadmap Matters
The WHO/Europe roadmap represents a shift from isolated AI initiatives toward coordinated infrastructure.
Its core model is:
Governance + Regulation + Data + Cybersecurity + Skills + Real-World Evaluation
For pharmaceutical companies, the message is relevant beyond health care delivery.
Many future pharmaceutical AI applications will depend on exactly this external ecosystem.
Regulatory Status Note
The Roadmap for action on artificial intelligence in health in the WHO European Region 2026–2033 was unanimously adopted by the 53 WHO European Region Member States on 29 October 2026.
It is a WHO regional policy roadmap.
It is not pharmaceutical GMP guidance, does not replace EU or national legislation, and does not create direct GMP validation requirements for pharmaceutical companies.
Its relevance to pharma lies primarily in digital health, clinical development, real-world evidence, pharmacovigilance and AI applications interacting with health-system data and infrastructure.
Sources
WHO/Europe – Governments across Europe and Central Asia unanimously adopt new vision for AI in health care, published 29 October 2026:
https://www.who.int/ … or-ai-in-health-care
WHO – Roadmap for action on artificial intelligence in health in the WHO European Region 2026–2033:
https://iris.who.int/handle/10665/387193
WHO – Roadmap implementation support guide: