UK Accepts All 44 Recommendations on Regulating AI in Healthcare
The UK government said on October 6, 2026 that it accepts all 44 recommendations of an independent commission on regulating AI in healthcare. Most details are due by spring 2027.
// ai in medicine
AI in Medicine Regulation Tracker
A sourced timeline of how regulators have handled AI in clinical care since 2023. It covers FDA guidance and device decisions, EU and UK rules, WHO guidance and U.S. payment policy, each linked to an official document.
Researched and fact-checked by AI, with no human review. How we verify
28 entries from 32 sources. Checked for news weekly; last checked
What this tracker follows
This tracker lists regulatory milestones for artificial intelligence (AI) in clinical care from 2023 to October 6, 2026. It covers the U.S. Food and Drug Administration (FDA) and other U.S. health agencies. It also covers the European Union, the United Kingdom and the World Health Organization.
An entry is a dated action by a regulator, a public payer or the World Health Organization. Each one is documented on an official page. The action may be a guidance document, a rule or formal proposal, a market authorization or a payment policy. It may also be an official review and the government's response to it. The three device entries are examples, not a full record. Each is a De Novo grant. That is the FDA pathway for novel devices that have no legally marketed predicate. Each grant is for a product on the agency's list of AI-enabled medical devices. The FDA says that list is not comprehensive, and it includes other De Novo grants from the same period. Drafts and proposals are labeled as such and are not binding until finalized.
The tracker leaves out:
- routine clearances of devices similar to ones already on the market
- breakthrough designations announced only by companies
- U.S. state laws
- court cases
- advisory committee meetings
- AI used to develop drugs
- market forecasts
One scheduled date depends on pending legislation. A December 2025 European Commission proposal would alter how the EU AI Act applies to medical devices. On October 6, 2026, the European Parliament's Legislative Observatory listed that file as awaiting a committee decision.
Scheduled next
Targets set by the organizations named. Dates like these often move.
Scheduled
EU AI Act high-risk rules due to apply to AI in medical devices
Regulation (EU) 2026/1744 was signed on July 8, 2026. It sets this date for AI systems that are high-risk under Article 6(1) and Annex I of the AI Act. A recital names the EU medical device and in vitro diagnostic regulations among the laws listed in Section A of that annex.
Source: Regulation - EU - 2026/1744 - EN - EUR-Lex, Publications Office of the European Union (EUR-Lex)
Scheduled
UK MHRA intends to launch consultation on classifying software and AI devices by spring 2027
The UK government's October 6, 2026 response to the National Commission says the MHRA intends to launch a consultation by spring 2027. The consultation would cover how software and AI-enabled devices are qualified and classified. The MHRA would then use secondary legislation to update the definition of a medical device.
Scheduled
UK MHRA intends to issue draft guidance on AI device change control plans by December 2026
The UK government's October 6, 2026 response to the National Commission says the MHRA intends to issue draft guidance by December 2026, for public comment. The guidance would cover predetermined change control plans. It would describe a proposed approach that lets manufacturers set boundaries for allowable change.
Timeline
UK government accepts all 44 recommendations on regulating AI in healthcare
The UK government accepted all 44 recommendations of the National Commission into the Regulation of AI in Healthcare. It also set out its first delivery priorities and next steps. It did so in a policy paper published by the Medicines and Healthcare products Regulatory Agency (MHRA).
UK National Commission publishes recommendations on regulating health AI
The independent commission was set up by the MHRA in September 2025. It recommended staged authorizations for new AI models, which the announcement compared to L-plates for learner drivers. It also recommended continuous real-world monitoring of AI-enabled devices. The MHRA said it heard from more than 12,000 people.
FDA issues discussion paper on regulating generative AI medical devices
The FDA said the paper outlines a possible two-axis framework for assessing risk. It said the paper also outlines premarket evaluation that combines non-clinical device benchmarking with clinical confirmation, and risk-proportionate postmarket monitoring. It asked for feedback under docket FDA-2026-N-7874 by October 19, 2026.
UK MHRA clarifies when AI scribes are medical devices in Great Britain
Guidance developed with NHS England says some ambient voice technology products are not regulated as medical devices. It says these are products intended solely to transcribe or summarize clinical conversations, draft letters or suggest clinical codes for clinician review. It says products intended to support diagnosis or treatment are regulated.
EU AI Omnibus enters into force, delaying high-risk AI rules
The European Commission said the AI Omnibus, proposed on November 19, 2025, entered into force on this date. It said rules for high-risk AI embedded in physical products covered by Annex I of the AI Act apply from August 2, 2028. It said rules for Annex III systems apply from December 2, 2027.
CMS proposes interim Medicare outpatient payment policy for clinical software
In its 2027 hospital outpatient proposed rule, the Centers for Medicare & Medicaid Services proposed a new term. Software as a Medical Service would replace the agency's term Software as a Service. The agency proposed applying it to 36 billing codes and assigning them to new technology payment groups. Comments were due August 31, 2026.
FDA issues revised Clinical Decision Support Software guidance
The guidance, issued January 29, 2026, supersedes a version issued January 6, 2026. It sets out which clinical decision support functions fall outside the legal definition of a device under the 21st Century Cures Act. It says existing digital health policies still apply to software that is a device.
Medicare's WISeR model starts testing AI-assisted reviews of selected services
CMS says the Wasteful and Inappropriate Service Reduction model runs from January 1, 2026 to December 31, 2031. It says the model runs in New Jersey, Ohio, Oklahoma, Texas, Arizona and Washington. The model pairs technologies such as AI and machine learning with human clinical review. CMS listed it as active in October 2026.
Source: WISeR (Wasteful and Inappropriate Service Reduction) Model, Centers for Medicare & Medicaid Services
HHS proposes removing AI model card requirements from health IT certification
A proposed rule from ASTP/ONC, the HHS health IT office, would narrow the decision support interventions criterion. That would remove what it calls AI model card requirements. It would delete all source attribute rules and the risk management rules for predictive tools. Comments were due February 27, 2026.
European Commission proposes changing how the AI Act covers medical devices
Proposal COM(2025) 1023 would revise the EU medical device and in vitro diagnostic regulations. It would also move both laws from Section A to Section B of Annex I of the AI Act. A recital says the aim is to prevent overlapping requirements for AI-enabled devices. It is a proposal, not law.
Source: EUR-Lex - 52025PC1023 - EN - EUR-Lex, Publications Office of the European Union (EUR-Lex)
FDA seeks comment on measuring real-world performance of AI-enabled devices
The FDA asked for information on best practices and methods for measuring and evaluating AI-enabled medical devices in real-world use. That included ways to identify and manage performance drift. It said it would consider comments submitted to docket FDA-2025-N-4203 by December 1, 2025.
EU publishes FAQ on how the AI Act and medical device rules interact
The European Commission's health directorate posted MDCG 2025-6, a set of frequently asked questions. It covers how the Medical Devices Regulation and the In Vitro Diagnostic Medical Devices Regulation interact with the Artificial Intelligence Act. The document is dated June 2025.
FDA grants De Novo request for Clairity's breast cancer risk software
The FDA granted Clairity's De Novo request for Allix5, received on September 6, 2024. The FDA database lists the device classification as radiological software device to predict future breast cancer risk, under regulation number 892.8500.
Source: Device Classification Under Section 513(f)(2)(De Novo), U.S. Food and Drug Administration
FDA issues draft guidance covering the full life cycle of AI-enabled devices
The draft recommends what developers should put in marketing submissions, including how they will monitor performance after launch and address transparency and bias. An FDA official said the agency had authorized more than 1,000 AI-enabled devices. The FDA still listed it as a draft in October 2026.
FDA finalizes guidance on predetermined change control plans for AI devices
The final guidance recommends that a predetermined change control plan describe planned modifications to an AI-enabled device. It says the plan should also cover how they will be developed and validated, and their impact. That way, each listed change does not need its own marketing submission. The FDA now lists a version dated August 2025.
EU AI Act enters into force with medical AI software in the high-risk class
The European Commission says high-risk AI systems include AI-based software intended for medical purposes. It says they must meet requirements on risk mitigation, data quality, user information and human oversight. Its page gives the original 36-month timetable for AI in regulated products, changed in 2026.
Source: Artificial Intelligence in healthcare, European Commission
UK MHRA launches AI Airlock regulatory sandbox for AI medical devices
The MHRA launched AI Airlock as a pilot regulatory sandbox for AI as a medical device. It said the pilot would start with four to six virtual or real-world projects, run through simulation. It said these would test regulatory issues that arise when such devices are used for direct clinical purposes in the NHS.
HHS nondiscrimination rule extends to patient care decision support tools
A final rule under Section 1557 of the Affordable Care Act bars covered entities from discriminating through patient care decision support tools, automated or not. It also requires reasonable efforts to identify and mitigate that risk. It took effect July 5, 2024. Those duties applied within 300 days.
FDA grants De Novo request for Prenosis sepsis software
The FDA granted the De Novo request from Prenosis for Sepsis ImmunoScore, received on May 5, 2023. The FDA database lists the device classification as software device to aid in the prediction or diagnosis of sepsis, under regulation number 880.6316.
Source: Device Classification Under Section 513(f)(2)(De Novo), U.S. Food and Drug Administration
WHO issues guidance on large multi-modal AI models in health
The World Health Organization released guidance on the ethics and governance of large multi-modal models. These are a type of generative AI that can take inputs such as text, images and video. WHO said the guidance sets out more than 40 recommendations for governments, technology companies and health care providers.
HHS health IT rule adds transparency requirements for predictive algorithms
The HTI-1 final rule revised the certification criterion for decision support interventions. Certified health IT must support source attributes, meaning categories of technical performance and quality information, for evidence-based and predictive tools. The rule took effect February 8, 2024.
FDA, Health Canada and MHRA publish principles for AI device change plans
The three regulators jointly identified five guiding principles for predetermined change control plans in machine learning-enabled medical devices. The MHRA describes such plans as a new way to manage the rapid changes common in software and AI products.
WHO publishes regulatory considerations on AI for health
WHO said the publication outlines six areas for regulating AI for health. It said they include transparency and documentation, risk management, data quality, and privacy and data protection. It said the aim is to set out principles that governments and regulators can use to write or adapt guidance on AI.
FDA grants De Novo request for Viz.ai machine learning cardiovascular software
The FDA granted Viz.ai's De Novo request for Viz HCM, received on January 10, 2023. The FDA database lists the device classification as cardiovascular machine learning-based notification software, under regulation number 870.2380.
Source: Device Classification Under Section 513(f)(2)(De Novo), U.S. Food and Drug Administration
FDA publishes draft guidance on change control plans for AI/ML devices
A Federal Register notice announced FDA draft guidance on predetermined change control plans for machine learning-enabled device software. Such a plan describes anticipated modifications so the agency can review them without a new marketing submission for each. Comments were due July 3, 2023.
Additional sources
- De Novo Classification Request, U.S. Food and Drug Administration
- List of Artificial Intelligence-Enabled Medical Devices, U.S. Food and Drug Administration
- Procedure File: 2025/0404(COD), European Parliament Legislative Observatory
- Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions, U.S. Food and Drug Administration
- Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations, U.S. Food and Drug Administration
Update history
- Rewritten in shorter, plainer sentences. No facts were changed.
- Added the MHRA's July 2026 guidance on AI scribes and two targets the UK government set on October 6, 2026, for draft guidance by December 2026 and a consultation by spring 2027.
Articles on AI in Medicine
Can AI Diagnose Disease? What It Does Well and Where It Fails
AI performs well on narrow, tested tasks such as reading eye and breast images, but 2026 studies found chatbots unreliable for assessing symptoms and urgency.
How Doctors Actually Use AI Today
In a 2026 AMA survey, most US physicians said they use AI, mainly to summarize research and draft notes. Trials of AI scribes show modest time savings, and hospital adoption is uneven.