United Kingdom - AI Action Plan

Overview

The "National AI Strategy - AI Action Plan" (published 18 July 2022) is the UK government’s cross‑departmental delivery update that operationalises the National AI Strategy (published 22 September 2021). The Action Plan presents progress and near‑term activity across the Strategy’s three pillars: long‑term investment in the AI ecosystem, diffusion of AI benefits across sectors and regions, and creating a governance framework to foster trustworthy innovation. It brings together departmental commitments, funded programmes and coordination mechanisms designed to accelerate R&D, expand skills pipelines and support public sector AI adoption while emphasising a risk‑based, context‑sensitive regulatory stance. The Action Plan is available from the government publication page: National AI Strategy - AI Action Plan (GOV.UK) and accompanies the National AI Strategy: National AI Strategy (Sept 2021) (GOV.UK).

Definitions

For the purposes of the Action Plan the government uses functional definitions rather than introducing new legal terms. "Artificial intelligence" is referenced as a family of technologies and techniques used to perform tasks that would otherwise require human intelligence. The Action Plan treats AI in a use‑based manner (focus on context and impact) and applies a risk‑based regulatory posture: high impact or high harm uses require stronger assurance and oversight. The Plan relies on existing statutory definitions where relevant (for example, data protection definitions under the UK GDPR / Data Protection Act 2018) and on cross‑government guidance such as the Cabinet Office algorithmic transparency standard and the Office for AI guidance materials.

Governance and Institutional Framework

The Action Plan sets out a coordinated institutional approach anchored by the Office for Artificial Intelligence and the Department for Science, Innovation & Technology, working with DCMS, BEIS (now departments reorganised in subsequent years) and other departmental leads. It emphasises strengthening regulators’ capabilities and asking sector regulators to take proportionate, context‑sensitive action under their existing statutory powers. The Plan references cross‑government initiatives such as the AI Assurance Roadmap produced by the Centre for Data Ethics and Innovation and the Cabinet Office’s national algorithmic transparency standard pilots to support public sector transparency. Responsible bodies and convenors are identified for major programmes, and the Plan commits to annual updates and baseline metrics to monitor progress. Relevant official pages include the Office for AI: Office for Artificial Intelligence (GOV.UK) and the pro‑innovation policy statement: Establishing a pro‑innovation approach to regulating AI (GOV.UK).

Key Focus Areas

The Action Plan organises delivery across interlocking themes: skills and talent (masters scholarships, Turing fellowships, conversion courses and visa measures), compute and data infrastructure (reviews and investments to secure compute, Trusted Research Environments and reforms to R&D tax reliefs to include cloud/compute costs), innovation funding and commercialisation (Innovate UK programmes, sector pilot funding), public sector adoption and procurement (AI Marketplace, public sector pilots and procurement guidance), standards and assurance (AI Assurance Roadmap, algorithmic transparency pilots, support for standards bodies), and governance/regulation (a cross‑sectoral, risk‑based approach that asks regulators to focus on actual harms and high‑risk use cases). The Plan highlights sectoral work (NHS AI Lab activity, Defence AI Strategy, MHRA work on Software and AI as a Medical Device) and identifies cross‑cutting priorities: fairness, safety, explainability, assurance and international collaboration. It signals that many aspects of AI will continue to be governed through existing sectoral statutes and regulatory regimes while the government supports tools and guidance to enable proportionate oversight.

Implementation Framework

The Action Plan is explicitly an implementation and delivery document. It sets departmental leads, milestone windows (near‑term 12 months actions), and funding commitments where available. Implementation mechanisms include: (i) targeted funding programmes (scholarships, Innovate UK calls), (ii) regulatory coordination (working with the Digital Regulation Cooperation Forum and sector regulators), (iii) standards and assurance development (supporting ISO/IEC standards and national pilot standards), and (iv) evidence and metrics (committing to develop baseline data on the size and scale of the UK AI sector and to publish annual progress updates). The Plan also commits to building the AI assurance ecosystem — including piloting assurance frameworks and supporting third‑party auditors and assurance providers — and to working with international partners to align approaches. For specific policy and delivery detail see the Action Plan page: AI Action Plan (GOV.UK).

Monitoring and Evaluation

The Government committed in the Action Plan to annual updates that track delivery against the Strategy’s vision. Monitoring focuses on a mix of quantitative and qualitative indicators: investment levels, skills pipeline metrics (doctoral training and conversion scholarships), adoption indicators in the public sector, progress on compute and trusted research environments, and delivery of regulatory and standards milestones. The Plan commits to developing new baseline data on the UK AI sector and to improving evidence collection to permit future iterations of the Action Plan to report progress. The government also signalled intent to work with independent bodies (Turing Institute, UKRI) to produce evidence and to consult regulators about appropriate metrics for measuring assurance and safety outcomes.

Penalties, Liability, and Appeals

The Action Plan is a policy document and does not itself create new statutory penalties. It reaffirms that compliance and enforcement in AI‑relevant domains continue to be exercised by existing statutory regulators under their legal powers. For example, breaches of data protection obligations involving AI processing are enforced by the Information Commissioner’s Office under UK GDPR and the Data Protection Act 2018, which can impose fines and corrective orders; safety failures in healthcare AI may be subject to MHRA regulation and civil liability; consumer protection or financial services breaches may attract action by the CMA or FCA. The Plan therefore clarifies responsibilities and encourages regulators to use proportionate enforcement, but it does not replace existing appeal routes or liability frameworks in primary legislation.

Relationship to Other Instruments

The Action Plan explicitly links to and builds on the National AI Strategy (Sept 2021), the pro‑innovation policy statement on AI regulation and other cross‑government documents such as the National Data Strategy. It situates the Action Plan alongside sectoral regulatory frameworks (Data Protection Act 2018 / UK GDPR, Consumer Rights, Health and Safety law, MHRA medical device regulation, National Security and Investment Act) and standards work (ISO/IEC and UK national standards). The Plan also references the Cabinet Office algorithmic transparency standard and the Centre for Data Ethics and Innovation outputs (AI Assurance Roadmap) as practical instruments to be used by departments and regulators.

International Alignment

The Action Plan emphasises international engagement: the UK seeks to shape global norms through participation in OECD, G7, the Global Partnership on AI and standardisation bodies. It promotes alignment where possible while protecting a pro‑innovation domestic stance; the Plan explicitly signals interest in interoperable assurance and standards regimes to facilitate trade and cooperation. The government also highlights bilateral research and R&D cooperation (for example the US‑UK AI R&D declaration) and pledges to continue advocating for the UK approach in international fora. See: Pro‑innovation approach (GOV.UK).

Implementation Timeline

Sources and References