EU GMP Annex 22: what the first AI annex means for your GxP practice

What this is about: the first GMP annex dedicated to AI

Until 2025 there was no place in the EU GMP guide where the use of artificial intelligence was regulated in its own right. AI-assisted systems ran under Annex 11 (Computerised Systems) — a framework written long before machine-learning models. That is changing: on 7 July 2025 the European Commission published the draft of a new Annex 22 “Artificial Intelligence” for targeted consultation — together with a draft revision of Annex 11. The consultation window closed in early October 2025. How the two relate, in Annex 22 vs Annex 11.

With it, AI in the GMP environment gets an explicit set of expectations for the first time. That is good news for everyone who wants to use AI in a controlled way: instead of uncertainty (“is this even allowed?”) there are now named requirements against which a deployment can be designed and checked.

Important for context: at the time of writing, Annex 22 existed as a draft; the final version may change in detail. The process is active: the EMA is still evaluating the consultation feedback and convened a multistakeholder workshop on the further development of Annex 22 for 30 June and 1 July 2026 — to gather expert contributions on control and mitigation measures such as guardrails for a risk-based approach. According to the EMA, the 2025 consultation feedback showed support for potentially enabling technologies such as generative AI and LLMs in medicines manufacturing; the EMA has not yet named a date for the final version. The control disciplines named in the draft — risk-based, data-disciplined and under human oversight — are already useful preparation. Always check the current status of the document before making decisions.

Scope: which AI Annex 22 means — and which it doesn't

The draft deliberately keeps the scope narrow. It addresses AI/ML models in critical applications of GMP-regulated manufacturing — wherever model output can directly touch product quality, patient safety or data integrity.

Three boundary lines matter most:

• Static models: for critical applications the draft expects models with deterministic behaviour — the same input leads to the same output. The model is trained, frozen, tested and then operated in a defined state.
• Dynamic models: systems that keep learning in operation and continuously change their behaviour are not foreseen for critical applications — their validated state could not be demonstrated as stable.
• Generative AI and LLMs: because of their probabilistic behaviour they sit outside the critical Annex 22 scope — the draft does not foresee their use in critical GMP applications. In non-critical, supporting applications they remain possible — then adequately qualified and trained personnel must review the outputs and remain responsible for their suitability for the intended use.

The draft does not foresee generative AI in critical GMP applications. In non-critical use, qualified personnel remain responsible for output suitability.

The core requirements at a glance

The draft organises its requirements into named chapters — among them Scope, Principles (with Documentation and Quality Risk Management), Intended Use (including Human-in-the-loop), Acceptance Criteria, Test Data, Test Data Independency, Test Execution, Explainability, Confidence and Operation. Thematically these condense into seven recurring disciplines — every validation team already knows each of them in spirit; what is new is their consistent application to models:

• Intended use: the model's purpose is precisely described — task, limits, input data, affected processes.
• Data quality & governance: training, validation and test data are controlled, representative and traceably managed.
• Independent test data: performance is assessed on data that was not used in training — the separation is demonstrable.
• Performance & acceptance criteria: metrics and thresholds are fixed before testing and aligned with the intended use.
• Explainability & confidence: where possible, it becomes visible which features drive a result and how confident the model is.
• Human oversight (in the draft: Human-in-the-loop): human supervision is built into the process — with defined roles and documented decisions.
• Monitoring & change control: model performance is monitored in operation (drift); changes run under control.

What this means concretely for QA and validation teams

Even with the final version pending: anyone using or planning AI in GxP processes can do four things right away.

First: build an inventory. Which AI runs in your processes today — including unofficially? A copilot in document drafting is AI use, even if it appears in no system register. Without an inventory there is no risk assessment.

Second: classify criticality. Does an application separate critical decisions (approval, specification, assessment) from supporting work (draft, research, structuring)? The draft draws its line exactly there.

Third: build human oversight as a process, not a claim. “A human looks over it” is not enough. What holds up: defined review steps, attributable review decisions with a reason, and an audit trail that keeps AI suggestion and human decision distinguishable.

Fourth: establish source binding. When AI output flows into regulated documents, it must be traceable what every statement rests on. A draft without verifiable sources is more expensive in review than no draft at all.

Annex 22, GAMP 5 2nd Edition, FDA CSA: one coherent picture

Annex 22 does not stand alone. Three frameworks from different directions have converged on the same principles in recent years:

• GAMP 5 2nd Edition (2022) accepts AI-assisted work within the risk-based lifecycle and demands critical thinking instead of template documentation.
• FDA CSA (final guidance 2025) shifts effort from documenting to defensible reasoning — test depth follows risk.
• EU GMP Annex 22 (draft 2025) formulates the same logic AI-specifically for the first time: controlled data, demonstrable model behaviour, human oversight.

For your strategy this means: you do not have to react to three rulebooks separately. If you build your AI work on source binding, traceable human review and a complete audit trail, you create one control base that remains useful even if individual requirements change in the final Annex 22 text.

How to prepare today — without waiting for the final version

A realistic preparation path in four steps:

• 01 · AI inventory and criticality map — capture every AI touchpoint and classify each application: critical / supporting.
• 02 · Define the review gate — for every AI-assisted workflow, fix who reviews, what is reviewed and how the decision is documented.
• 03 · Control the source and data space — define which approved sources AI drafts may be built from, and secure the separation of working data and training data contractually and technically.
• 04 · Carry evidence from the start — versions, review decisions and reasons are created in the workflow, not in a re-documentation loop before the audit.

That is exactly the pattern traqx is built on: drafts are created from the controlled project sources, every regulatory statement ends on a clickable citation, every source reference is checked deterministically, and every human review decision remains connected to its person, status and audit trail. That is not an Annex 22 certificate — no such certification exists. It is a way of working that puts the draft's principles into practice today.

Frequently asked questions

What does EU GMP Annex 22 regulate?

The Annex 22 draft sets out requirements for the controlled use of AI models in critical applications of GMP-regulated manufacturing. It is the first annex of the EU GMP guide dedicated entirely to artificial intelligence; the European Commission published it as a draft for consultation on 7 July 2025, together with a draft revision of Annex 11. The draft sets out concrete expectations for controlled data, demonstrable model behaviour and consistent human oversight, against which an AI deployment can be designed and checked.

What does Annex 22 require for AI models?

The draft requires every model to have a precisely described intended use and its training, validation and test data to be controlled and traceably managed. Performance is assessed on independent test data against acceptance criteria fixed in advance, and where possible it becomes explainable which features drive a result and how confident the model is. Human oversight is built in as a process with defined roles, and model performance is monitored in operation — with drift in view and changes under control.

Which AI systems does Annex 22 apply to?

The draft applies to AI/ML models in critical GMP applications — wherever the output can directly touch product quality, patient safety or data integrity. For such critical applications the draft expects static, deterministically operated models: trained, frozen, tested and run in a defined state. Systems that keep learning in operation, as well as generative AI and LLMs, are not foreseen for critical applications; in non-critical, supporting applications AI remains possible when adequately qualified and trained personnel remain responsible for output suitability.

Is EU GMP Annex 22 already in force?

Annex 22 is not yet in force — it exists as a draft. The European Commission published it for targeted consultation on 7 July 2025, with the window closing in early October 2025; the final version was still pending at the time of writing. The EMA is still evaluating the feedback and convened a multistakeholder workshop on the further development of Annex 22 for 30 June and 1 July 2026; the EMA has not yet named a date for the final version. The direction is clearly discernible, but individual requirements — such as the future treatment of generative AI — may still change in the final text, so always check the current status of the document before making decisions.