What the MHRA's 2030 strategy says about AI, and which parts have a date

The Medicines and Healthcare products Regulatory Agency (MHRA) published its strategy to 2030 on Wednesday 14 October 2026. On our count, the only AI deliverable it commits to publishing in a set year is an AI Medicines Development Action Plan "in 2027". Its framework for AI as a medical device sits under a 2030 aim, with no delivery date.

The four AI headline commitments

The MHRA's press release lists four commitments under "Transform regulation with AI and pioneer MedTech methodologies":

• "Develop an AI-enabled Clinical Trial Protocol Checker to give sponsors regulatory certainty and support faster, higher-quality clinical trial applications."
• "Publish an AI Medicines Development Action Plan in 2027 setting out how AI will be used to transform medicines development."
• "Establish an MHRA Centre for AI and Digital Health to make clearer regulatory expectations for developers of digital and AI-driven medical devices and medicines."
• "Create a modern, adaptive regulatory framework for AI as a medical device."

The full strategy adds detail to each. The protocol checker is part of the agency's ARISE programme, and the strategy says it "will allow sponsors to assess their protocols against common Grounds for Non-Acceptance prior to submission". The Centre "will bring together specialist leadership and expertise within the MHRA to support the regulation of digital and AI-driven products, including medical devices and medicines." The strategy gives neither a delivery date.

Dated, firm, or exploratory

On our reading, the AI-related items, plus one related Yellow Card change, sort into three groups.

With a date

• The AI Medicines Development Action Plan: "the MHRA will publish an AI Medicines Development Action Plan in 2027."
• Yellow Card (not described as AI): "In 2027, the MHRA will begin to introduce a more conversational interface for Yellow Card reporting". The strategy does not say what technology will power it.
• The overall goal: "By 2030, the MHRA aims to deliver the global standard for safe and agile regulation of MedTech and AI in healthcare." That is an aim, not a deliverable.

Firm wording, no date

• "we will introduce expanded Predetermined Change Control Plans, enabling pre-agreed model updates to be implemented without unnecessary delay."
• The Centre for AI and Digital Health, and the protocol checker.
• "We will routinely use AI to reduce administrative burden". The strategy says the AI tools in its ARISE programme "will support and enhance, not replace, human expertise".

Exploratory or conditional

• "We will also explore the use of Master Files for foundation models".
• Regulatory sandboxes: the MHRA "will support proposed legislative reforms to create statutory mechanisms for regulatory sandboxes", with "Initial areas of focus" on "AI as a Medical Device, and AI in pre-clinical testing (ADMET)."

What it means for patients

The patient explainer mentions AI once, saying that as "AI-enabled medical devices" are developed the MHRA "will maintain high standards for safety and effectiveness while adapting how we assess and monitor them."

On safety monitoring, the strategy says the agency will use "advanced statistical, computational and AI-based approaches to detect emerging signals and help predict risks before they affect patients". It also says the MHRA will work against "mis- and dis-information about regulated products on social media and AI platforms", because "more people rely on social media, online content and AI tools for health information".

The patient explainer says the new Yellow Card interface will make it "simpler to report suspected problems with medicines and medical devices", and that the MHRA "will also explore integration with the NHS App".

What it means for anyone building health AI

The strategy describes AI as a medical device as presenting "changing performance across time, settings and populations". It says "The growing use of foundation models", which are "trained on large, broad datasets", further amplifies "the variability and complexity". Its answer is "continuous monitoring of real-world performance, supported by clearly defined accountability across developers, deployers and health system partners", and a whole-system "approach to safe deployment, combining developer responsibility, provider governance and professional oversight".

The MedTech explainer sets that work a target date: "By 2030, we aim to deliver a globally leading approach to regulating AI as a medical device."

For medicines developers, the strategy says the MHRA and the Regulatory Innovation Office are partnering on the "Beyond ADMET: AI for Medicines Safety" sandbox, "exploring how AI-generated evidence can be used responsibly".

Some dates sit outside this strategy. As we reported on 10 October, the government's 6 October response to the National Commission into the Regulation of AI in Healthcare said the MHRA "intends to issue draft guidance by December 2026" on change control plans. The strategy does not repeat that date.

The MedTech explainer also lists "staged authorisation" in its AI-as-a-medical-device approach, a phrase the full strategy does not use. The government response describes "novel staged authorisation pathways that enable earlier access while generating real world evidence", and says the MHRA "will explore options to introduce staged authorisation pathways". Our July piece covered what the MHRA's AI Airlock sandbox found when it tested clinical AI tools.

Why it matters

On our reading, the strategy is a direction of travel rather than a rulebook. The Centre and the foundation-model Master Files have no delivery dates in the strategy, and the device framework sits under a 2030 aim rather than a delivery date. For patients, one change with a date is a simpler Yellow Card interface, which the MHRA says it will begin introducing in 2027. For builders, most of the detail is still to come.