What the AI Growth Lab actually is
On 21 October 2025 the Department for Science, Innovation and Technology (DSIT) published its Blueprint for AI Regulation, and the centrepiece of that blueprint is a proposal for an "AI Growth Lab", described by government as a pioneering cross-economy sandbox. The core mechanism is straightforward, government would temporarily modify or waive specific regulatory requirements for a limited period so that a licensed company can trial an AI product or service under supervision, in live market conditions, rather than in a lab-only test environment. The rationale set out in the accompanying call for evidence is that some existing rules were not written with AI systems in mind and can create uncertain or effectively prohibitive compliance obligations for firms trying to deploy AI responsibly, so a supervised, time-limited relaxation lets both the firm and the regulator learn what safe deployment actually looks like before any permanent rule change is made, according to the official call for evidence.
Two operating models were put out for consultation. Under Model 1, a single government-operated Lab would act as a unified entry point for applicants, working with sectoral regulators through an oversight committee. Under Model 2, a lead regulator would be appointed for each individual sandbox, based on their existing statutory remit, with consortium arrangements used where an application spans more than one regulator's territory, per the same call for evidence.
Which sectors and regulators are involved
The call for evidence names professional services, healthcare, transport and advanced manufacturing as priority sectors for early pilots, and gives illustrative examples including AI supporting environmental impact assessments in planning, autonomous AI in diagnostic imaging such as radiology within the NHS, and robotics or autonomous delivery devices navigating pavements and roads. It references existing sector regulators as the natural partners for sandbox oversight, pointing to the Medicines and Healthcare products Regulatory Agency, which already runs its own AI Airlock programme for AI as a medical device, the Financial Conduct Authority, which operates its own Innovate sandbox, the Bank of England and the Information Commissioner's Office, which operates a Data Protection Sandbox, according to the call for evidence.
How a firm would actually apply
The call for evidence set out that businesses could engage with the Lab by submitting an application through an online survey or by email, and that the government intended participation to be open to start-ups, established FTSE-listed companies and global AI developers alike, rather than being restricted to any one size or type of firm. Applications would be assessed against eligibility and safeguard criteria that were themselves subject to consultation, covering matters such as the intended length of a sandbox period and the safeguards a participant would need to demonstrate before being admitted, per the official page.
The consultation itself ran from 11:30am on 21 October 2025 to 11:59pm on 7 January 2026, per the gov.uk call for evidence page. The page carries an update notice dated 18 December 2025, and an earlier published version of the same call for evidence document had listed a 2 January 2026 closing date, so the deadline appears to have been pushed back in the final weeks of the consultation. Responses will inform how DSIT finalises the Lab's design, including which of the two operating models it adopts.
Safeguards and oversight during a sandbox trial
Government has been explicit that the Growth Lab is not intended as a vehicle for general deregulation. The call for evidence describes a set of protected areas, or "red lines", that would not be modified or waived under any sandbox arrangement, including consumer protections, safety provisions, fundamental rights, workers' protections and intellectual property rights. Separately, consumer redress, whether provided for under consumer protection legislation, in contract or in tort, is described as remaining unaffected, so participation in a sandbox would not remove a consumer's ability to complain or seek remedy, according to the call for evidence. The government's stated intent is that successful pilots feed into permanent reform, through updated regulator guidance, codes of practice or, where needed, statutory amendments, rather than the relaxation simply lapsing with no wider learning captured.
How this fits the UK's existing five-principles approach
The Growth Lab sits inside, rather than replaces, the framework DSIT set out in its March 2023 white paper, "AI Regulation: A Pro-Innovation Approach", published 29 March 2023 and updated 3 August 2023. That framework asks existing regulators, rather than any new AI-specific regulator or statute, to apply five cross-sector principles when supervising AI within their own remit, safety, security and robustness, appropriate transparency and explainability, fairness, accountability and governance, and contestability and redress, according to the white paper. The Growth Lab extends that same logic, using bodies such as the FCA, MHRA and ICO as the sandbox overseers rather than creating a new central AI authority.
One important qualification surfaced in parliamentary debate, in a House of Lords exchange on 26 March 2026, Baroness Lloyd of Effra, responding for the government, confirmed that giving the Growth Lab the legal power to actually waive statutory requirements will require primary legislation, a step that had not yet been taken by the time of that debate, per the Hansard record. Separately, a Private Member's Bill from Lord Holmes of Richmond, the Artificial Intelligence (Regulation) Bill, would go further and create a statutory AI Authority to oversee how existing regulators handle AI, but as of mid-2026 it has no government backing and the government's own approach remains non-statutory outside of the Growth Lab mechanism.
Current status as of July 2026
The full, rule-waiving version of the Growth Lab consulted on in the October 2025 call for evidence has not yet formally launched. What has launched is a narrower, advisory AI Growth Lab, announced by government on 8 June 2026 with legal services, specifically LawTech, legal service providers and conveyancing, as its first focus area. This advisory Lab brings together the Solicitors Regulation Authority, the Council for Licensed Conveyancers, the Information Commissioner's Office and the Legal Services Board to help AI innovators understand how existing rules apply to novel AI use cases, according to the gov.uk announcement. Applications for LawTech companies, legal service providers and conveyancing firms were expected to open later in summer 2026.
Government has been explicit that this advisory version is a different instrument from the full sandbox model, participation in the Lab does not constitute regulatory approval, endorsement or authorisation, and legal and regulatory requirements remain unchanged for participants, per the same announcement. The Information Commissioner's Office confirmed its role in the legal services Lab in a statement issued around the same launch date. Further sandboxes covering the other priority sectors named in the original consultation, healthcare, professional services, transport and advanced manufacturing, are expected to be stood up sector by sector through 2026 and 2027, but no firm launch dates for those had been published as of late July 2026. Firms in scope of a future sandbox sector should watch DSIT's consultation outcome publication and any primary legislation introduced to grant the Lab its rule-waiving powers, since that legislative step remains the gating item for the full model to operate as originally envisaged.
Primary sources: AI Growth Lab call for evidence, DSIT · Advisory AI Growth Lab for legal services, GOV.UK · AI Regulation: A Pro-Innovation Approach white paper, GOV.UK · House of Lords debate on the AI Growth Lab, Hansard
