What MHRA Strategy 2030 commits to on AI, and when

Checked against the strategy on October 20, 2026.

The UK Medicines and Healthcare products Regulatory Agency (MHRA) published MHRA Strategy 2030 on October 14, 2026. The strategy PDF is 40 pages long and runs to 2030. I count 19 separate AI items in it, including steps already taken. They range from safety signal detection to a new AI center inside the agency.

Only one of them is a dated deliverable, the AI Medicines Development Action Plan, due in 2027. A second date, 2030, belongs to an aim rather than a deliverable. Everything else either has no date or is already done. If you work in regulatory, quality, chemistry, manufacturing and controls (CMC), clinical or safety, 2027 is the only date you can put on the calendar. I would read the rest as a direction of travel.

Every AI item, dated or undated

Some rows are aims or plans to explore rather than firm commitments, and the wording in each row follows the strategy. Page numbers are the printed numbers in the PDF.

AI items in MHRA Strategy 2030The two dated items first, then the rest in page order. 19 items, one per row, with the printed page it is on. Last checked Oct 20, 2026.

Item	Page	Status
Publish an AI Medicines Development Action Plan	20	Dated, 2027
Aim to deliver “the global standard for safe and agile regulation of MedTech and AI in healthcare”	27	Aim, dated 2030
Routinely use AI to reduce administrative burden and improve the quality of decisions	11	Undated
Work under the ARISE program on where AI can safely automate parts of application handling, such as screening evidence and finding missing information	11	Undated, ongoing
Develop the AI-enabled Clinical Trial Protocol Checker	11	Undated
Use AI to identify and assess safety signals in large volumes of National Health Service (NHS) data	11	Undated
Use AI-based approaches to detect emerging safety signals and help predict risks	13	Undated
Work with partners on misinformation about regulated products on social media and AI platforms	16	Undated
Horizon scanning supported by data science and AI	17	Undated
Support proposed reforms for statutory sandboxes, with AI as a Medical Device and AI in preclinical testing (absorption, distribution, metabolism, excretion and toxicity, or ADMET) as the first areas	17	Undated
“Beyond ADMET: AI for Medicines Safety” sandbox with the Regulatory Innovation Office (RIO)	19	Undated, ongoing
AI considerations added to CMC regulatory frameworks, plus a synthetic data pilot in clinical research and work on AI in ADMET prediction	20	Already done, undated
Reform regulation to support emerging technologies, including AI	20	Undated
Use AI to support the review of early phase trial documentation	22 to 23	Undated, under a 2030 national aim
Expanded Predetermined Change Control Plans, and exploring Master Files for foundation models, for AI as a medical device	28	Undated
Establish an MHRA Centre for AI and Digital Health	28	Undated
Work with the Health AI Global Regulatory Network as a founding “pioneer” country	28	Undated
Explore AI automation to reduce administrative burden in human resources	33	Undated
Use the Leeds Digital Hub to “advance our digital, data and AI ambitions”	34	Undated

Source

MHRA Strategy 2030, published Oct 14, 2026 (printed page numbers)

Updated

Oct 20, 2026

Counted

Every item in the strategy that names AI

Record

mhra-strategy-2030-ai, table rev 1, Oct 20, 2026

The commitments pharma teams should read first

The AI Medicines Development Action Plan, in 2027

Chapter 05 says the MHRA “will publish an AI Medicines Development Action Plan in 2027.” The strategy says the plan “will set out a coordinated UK-wide approach, developed with partners including the RIO.” It gives a year but no month.

AI considerations in CMC frameworks, already done

The same section lists this as a step already taken: adding “AI considerations” into CMC regulatory frameworks. It names no specific guidance and gives no date, so CMC and quality teams should not read this as a new CMC AI guideline.

The AI-enabled Clinical Trial Protocol Checker, in development

The MHRA says it “is developing an AI-enabled Clinical Trial Protocol Checker that will allow sponsors to assess their protocols against common Grounds for Non-Acceptance prior to submission.” It is part of the ARISE program. In my view it is the most practical item in the strategy for clinical and regulatory teams. It is still in development and has no launch date, so sponsors can’t use it yet.

The MHRA Centre for AI and Digital Health, undated

Chapter 07 says the MHRA “will establish an MHRA Centre for AI and Digital Health.” It will bring together specialist expertise “within the MHRA” to support the regulation of digital and AI products, “including medical devices and medicines.” It is an internal MHRA unit with no start date.

What to do now

These are my own suggestions for pharma teams. The strategy does not require any of them.

1. Add the 2027 action plan to your regulatory intelligence watch list and check the gov.uk page from time to time.
2. If you use AI anywhere in manufacturing or quality, ask your CMC lead to confirm which MHRA CMC frameworks now include AI considerations.
3. Review the Grounds for Non-Acceptance in your recent UK trial applications. That review is useful now, and it prepares you for the protocol checker whenever it launches. I would not build a submission timeline around the checker.
4. If you work in safety, note that the MHRA plans to use AI to find signals in NHS data. In my view, that makes your own signal detection more important, because the agency may spot a signal in that data before you do.

The MHRA documents I track are on the Pharma AI regulatory guidance page of the Pharma AI Tracker.

---

Want to plan for MHRA AI expectations? That is the work I do.