The programme will explore whether AI can improve the assessment of accuracy and safety, better predict risks, and detect effects that existing approaches may not capture.
Adverse drug reactions lead to around 250,000 hospital admissions in the UK each year and cost the NHS more than £2bn annually.
Around 90 per cent of medicines fail during development, partly because existing methods cannot always predict how they will behave in people.
The government said uncertainty during early safety testing can also prevent potentially useful treatments from progressing.
The Medicines and Healthcare products Regulatory Agency will establish a new regulatory sandbox, giving companies and researchers a controlled testing environment to work with regulators on how AI tools could be used to better understand medicines safety and predict potential side effects.
A regulatory sandbox is a controlled environment where companies and researchers can test new technology with oversight and guidance from regulators.
It is being supported by funding from the government’s Regulatory Innovation Office.
The sandbox will allow innovators to test AI tools that could predict how medicines behave in the body, including how they are absorbed, processed and whether they may cause harm.
The programme will also explore how better use of clinical data can improve understanding of how medicines work across different groups.
This will include populations often underrepresented in studies, such as children, older people and people from diverse backgrounds.
The MHRA will use this work to understand how reliable these tools are and whether they can support decisions about the safety of new medicines.
The regulator said building a stronger evidence base and setting clear expectations for safe use could give companies confidence to invest in innovation in the UK.
The initiative forms part of wider efforts to modernise medicines development, including the use of advanced modelling and synthetic data to improve how safety and effectiveness are assessed.
Synthetic data is artificially generated information designed to reflect real data without directly identifying individual patients.
The programme will support the government’s AI for Science Mission One to develop new treatments faster.
It also reinforces plans announced in November to drive alternatives to animal testing.
The government said this also supports the 10 Year Health Plan ambition for the UK to have the world’s most AI-enabled healthcare system, helping patients get faster access to safe, effective and innovative treatments.
Up to five AI-driven approaches will be tested in the first phase.
The MHRA will begin working with industry and academic partners from summer 2026 to shape how the sandbox operates.
Health innovation minister Preet Gill said: “The AI revolution is here, and we want our NHS staff to be the first in the queue, armed with rigorously tested clinical AI tools.
“By giving innovators a safe space to test these tools alongside regulators, we can build the evidence base needed to get safer, more effective treatments to patients faster. That means fewer adverse reactions, less reliance on animal testing, and a smarter, more efficient medicines development process.
“Through our 10 Year Health Plan, we are driving the NHS to be the most AI-enabled healthcare system in the world.”
Science minister Lord Vallance said: “Too many promising medicines fail late in development or never reach patients because the evidence needed to support them is difficult and slow to generate.
“By leveraging our strengths in life sciences, AI and pro-innovation regulation, this sandbox will help make the UK one of the best places in the world to develop the next generation of medicines safely.”
Lawrence Tallon, chief executive of the MHRA, said: “We’re seeing extraordinary advances in AI and biomedical science. The opportunity now is to harness them to deliver real benefits for patients.
“These technologies could help us understand medicines better, generate stronger evidence on their safety, and accelerate the development of innovative treatments, especially in areas of unmet need.
“For patients, that means greater confidence that the medicines they rely on are supported by the best available science, with evidence that better reflects the diverse range of people they are intended to treat.
“By working directly with developers and researchers, we can help create the right environment for innovation to thrive and cement the UK’s position as a world leader in life sciences.”
Prof Alastair Denniston, NHS doctor and chair of the National Commission into the Regulation of AI in Healthcare, said: “This sandbox is a valuable opportunity for the UK to develop a practical, trusted approach to assessing AI models that predict drug safety and pharmacokinetics: helping regulators, researchers, and innovators understand what good evidence looks like, and supporting responsible innovation that accelerates the development of new medicines that will benefit patients.”
Pharmacokinetics is the study of how a medicine enters, moves through and leaves the body.
Prof Chris Molloy, chief executive of the BioIndustry Association, said: “We welcome this timely and important step which will accelerate medicines R&D in the UK and globally. UK innovators are developing AI-driven technologies and UK biotechs will adopt them. AI models have the potential to derisk drug development and deliver them to patients faster. But they need to be taught, tested and proven in a rigorous, safe space – which this sandbox delivers. We look forward to enabling a continued dialogue between MHRA and our members nationwide.”
Dan O’Connor, executive director of regulatory policy at the Association of the British Pharmaceutical Industry, said: “We welcome the growing focus on AI model development as an important step towards more predictive and efficient medicines development. This sandbox could be a step towards the ‘safe harbour’ that ABPI has called for, creating the trusted environment industry needs to develop and adopt regulatory-compliant AI models in medicines development.
“To succeed, this work will need meaningful industry engagement and should align closely with the work of the CERSIs to help build a coherent, trusted framework for innovation and regulatory adoption. We look forward to being part of the discussions to move this forward.”
