AI Airlock Background
The AI Airlock was established to proactively investigate the regulatory challenges of AI as a Medical Device. Working with key partners during the pilot phases, the programme is working towards key project outputs including an improved regulatory guidance for AIaMD.
You can read more on the background here and on the initial phase in this post. Information on Phase 2 is available in this post.
AI Airlock applications
AI Airlock phase 3 call for applications
Candidates can now apply to the MHRA) to join the AI Airlock.
Please submit applications using this portal – Application Portal
- This application process is designed to gather information about candidates (including their AI enabled Medical Devices and prototypes, if they are a manufacturer) and evaluate their suitability for the AI Airlock sandbox.
- The application window will remain open year-round, and applicants may submit an application at any time until this page is updated to indicate that the call has been closed.
- Applications will be evaluated in scheduled review waves and onboarded on a rolling basis. Applications submitted after a scheduled review will generally be carried forward to the next review wave.
- For applicants that would like to be considered for the current application review wave, please apply before the 31st October 2026.
- The application window will remain open after this time and a subsequent shortlisting period will take place in Spring 2027 .
What is the current areas of focus?
The current priority regulatory challenge area is Post-Market Surveillance (PMS) and oversight across the lifecycle for AI enabled Medical Devices.
Specific focus will be given to projects that address the priority topics set out below:
- Ongoing assurance across lifecycle
- PMS that accounts for varying levels of local oversight
- Communicating information across the system effectively and throughout the lifecycle.
Further detail on each of the above is available on this page.
If you see a challenge area you are interested in exploring or addressing with the programme, and you have sufficient information to support hypothesis development and testing, complete the relevant section of the form.
- The MHRA are also open to expressions of interest from outside of these key focus areas. If there is a challenge you’d like to explore within the AI Airlock which does not fit into the PMS and lifecycle oversight challenge area, we would encourage you to submit your proposal using the form above and selecting the “Express interest in another regulatory focus area” option in the form.
- These expressions will help to inform future priorities for the programme which we will update in due course, however, please note that priority will be given to proposals that align with current areas of focus.
How will candidates included in the programme benefit?
During the Airlock programme candidates will benefit from a unique collaboration of industry and regulatory experts, a bespoke testing plan and gain an improved understanding of the current regulatory framework.
Participation in the regulatory sandbox presents an opportunity for a proactive approach to AIaMD regulation, derisk the innovative products ahead of entering the market and may benefit candidate’s future routes to market.
What will outputs of the AIrlock programme include?
Outputs of the Airlock programme will include:
- project reports from each candidate team
- an Airlock Sandbox report of learnings to inform future guidance and implications for the regulatory framework
- a programme evaluation report sharing learnings on the use of regulatory sandboxes.
What is the application process?
Candidates should complete the application form to submit a proposal to the AI Airlock. You can also view an MS Word copy of the application form here.
- When preparing answers, please note character limits may apply for each section.
- Organisations of all sizes are invited to apply.
- A lead applicant may partner with other companies or organisations.
- Following applications, candidates may be invited to a follow up shortlisting meeting, to share more information about their application, product, and regulatory challenge to be tested.
- Confidentiality and commitment agreements will be established between the Airlock programme and the participating organisations.
- Candidates may also be asked to share more information about their products or prototypes, where available, including:
- Evidence to demonstrate the basic principle of your idea.
- Evidence of relevant certification, regulatory and technical/clinical documentation.
- Details of a data access source, as relevant.
What are the eligibility criteria?
Candidates will be asked to commit to working with the AI Airlock programme for the duration of a bespoke testing plan that will be co-developed with the Airlock team during onboarding. The following criteria will also be considered during application sifting.
- The product or solution has potential to deliver benefits for patients
- Applicants can demonstrate that the product (or prototype) will provide a benefit to patients, the wider health system and public health.
- The specific pathway, population or health system challenge that will be impacted by the solution should be identified.
- The product or concept application is innovative
- The Airlock programme is looking to recruit projects that have the potential to be transformative in the healthcare and regulatory systems, interrogate current approaches and offer solutions to existing challenges, and align with the vision set by the National Commission into the Regulation of Healthcare in AI.
- The product or concept application addresses a regulatory challenge
- The Airlock provides a unique opportunity for products/ prototypes that provide solutions to regulatory challenges experienced by Artificial Intelligence enabled Medical Device products across the medical device lifecycle.
- The proposal is ready to be trialled and benefits from the AI Airlock environment.
- Applicants should be able to establish a testing plan for their project or product within the Airlock and commit resource to carrying out such plan.
- This should include a clear challenge, hypothesis, or goal to be addressed, high level objectives and testing plans, and outline relevant data access needs.
In collaboration with the Airlock programme partners, the Airlock team will take a portfolio approach to selecting the final products for the third phase against these criteria.
While no product will be selected that does not fulfil the relevant eligibility criteria, the MHRA will be selecting innovative products that will enable the AI Airlock programme to maximise the breadth of learning and potential impact of the phase.
Based on demand, the programme may also be able to establish a reserve list of candidates.
About the AI Airlock testing environments
During the Airlock programme innovators will benefit from a unique collaboration of industry and regulatory experts, co-development of a bespoke testing plan, and gain an improved understanding of the current regulatory framework.
- Participation in the regulatory sandbox presents an opportunity for a proactive approach to refine AI regulation, derisk innovative products ahead of seeking market entry, and generate evidence and solutions that can inform future national policy.
- Each Airlock project will be designed specifically around tangible outputs such as:
- regulatory tools
- frameworks or
- best practices
- These outputs are intended to be made public for the benefit of the wider community of regulators, innovators and other relevant stakeholders.
- Further, targeted outputs may be accompanied by project reports from candidate teams, further Airlock sandbox reports to outline learnings and possible implications for the regulatory framework, and programme evaluation reports to improve future work.
The AI Airlock has been designed to explore regulatory challenges within three different “airlock environments” that can be utilised as part of a bespoke sandbox testing plan. They are:
“Simulation airlock”
- This is a roundtable workshop established to bring together key stakeholders and experts to address relevant focused questions, to gather perspectives and to move towards recommendations and regulatory solutions in a safe space.
- This environment can be used for challenge areas that need cross-functional alignment / multi-stakeholder engagement or for conceptual products in early stages of development or where data access is not available or needed.
“Virtual airlock”
- This is a digital testing space built into or around an AI system, where its functions can be explored safely.
- It allows testers to see how the AI performs in a virtual environment using real or synthetic data, feeding it information, checking how it responds, and inspecting how its internal processes work.
- It is designed to simulate real tasks, so teams can evaluate things like accuracy, reliability, and reasoning within the AI itself.
Real-World airlock”
- This involves observing the device or solution in the actual environment where the AI product would eventually be used (but run safely and outside patient pathways (e.g. shadow deployment).
- The tool is used alongside doctors and real (retrospective) patient data, following the usual clinical processes, but its outputs are not used to make real decisions about patients.
- The goal is to see how the AI performs in realistic practice under non-ideal conditions: how it may fit into workflows, how accurate or helpful it is, and how it can be demonstrated to be safe and reliable in actual use.
These environments can be utilised in different ways depending on the challenges to be addressed as part of a testing plan developed with candidates during onboarding. Applicants are invited to outline their own access to virtual or real-world environments and data if needed for testing in the Airlock.
Who are the AI Airlock partners?
The AI Airlock Partners comprise of the following representatives:
- MHRA
- Department of Health and Social Care (DHSC)
- AI Airlock Community of Experts, including representatives across government,
- Approved Bodies, industry, clinicians, patient representatives and sector regulators .
The Airlock works across government departments to deliver its activities including with the Regulatory Innovation Office.
The Information Commissioner’s Office (ICO) will also be supporting the MHRA AI Airlock via a referral service offering data protection by design advice to applicants. If you would like support, please indicate this during the application process.
About Fees & Funding
Is there a fee for application or participation in the programme?
There is no fee for application or participation in the AI Airlock programme.
The selection process and advice provided to applicants selected for the cohort from the consortium of the Airlock Partners will be free of charge.
Resource commitment between candidates and the MHRA
While there is no fee for applicants throughout the programme, there will be a resource commitment to be made between the candidates and the MHRA upon joining the Airlock. Currently, candidates are expected to fund their own studies and delivery of any Airlock testing, including accessing relevant data sets.
What are the phase 3 testing timelines?
The current phase of the AI Airlock is funded until April 2029.
While each sandbox testing plan will be bespoke to the product and challenge, candidates should generally expect to scope their individual Airlock testing/engagement to within 6 – 12 months.
Source: MHRA