UK publishes policy paper that sets out an ambitious ten-year plan to remain a global AI power

On 18 July 2022, the UK Government published the policy paper 'Establishing a pro-innovation approach to regulating AI', which headlines an approach that steers clear of AI legislation in the UK. This represents a significant departure from the EU's approach which has been driving a comprehensive legislative agenda for AI. The proposed "light touch" approach will be welcomed by those in the AI industry who feel that onerous regulations can hold back innovation. However, in an increasingly global world, many companies will be required to comply with the EU regime in any event, so the benefits of the UK's approach may be limited. However, it remains unclear how the proposal impacts the plans of the Office for Product Safety & Standards (OPSS), the UK product safety regulator, to amend the product safety and product liability framework to address new technologies including AI.

Establishing pro-innovation, clear and flexible approaches

The policy paper recognises that there are challenges facing AI businesses, including a lack of clarity as to what law applies, overlaps in regulation and inconsistency in approach in different sectors. In an attempt to remedy these challenges, the paper sets out an ambitious ten-year plan for the UK to remain a global AI power that will be grounded in a set of cross-sectoral principles, which are:

  • to ensure that AI is used safely;
  • to ensure that AI is technically secure and functions as designed;
  • to make sure that AI is appropriately transparent and explainable, e.g., requiring information to be provided for the data being used, relating to training data, the logic and process used, information to support the 'explainability' of decision making and outcomes by the AI product, to name a few;
  • to embed considerations of fairness in AI;
  • to define legal persons' responsibility for AI governance, i.e., accountability for the outcomes produced by AI must rest with an identified/identifiable legal person; and
  • to clarify routes to redress or contestability relating to users' ability to contest an outcome.

The principles must be:

  • Context-specific - regulations should be based on the use and impact of the AI technology and to delegate responsibility for designing and implementing proportionate regulatory responses to regulators (e.g. Ofcom, the Competition and Markets Authority, the Information Commissioner's Office, the Financial Conduct Authority, and the Medicine and Healthcare Products Regulatory Agency).
  • Pro-innovation and risk-based - regulators should focus on high-risk concerns rather than hypothetical or low risks associated with AI.
  • Coherent - cross-sectoral principles should be tailored to the distinct characteristics of AI, and regulators should interpret, prioritise and implement these principles within their sectors and domains.
  • Proportionate and adaptable - cross sectoral principles should be set on a non-statutory basis to remain adaptable, and regulators should consider "lighter touch options" such as issuing guidance or voluntary measures.

The principles track the EU's approach in its draft AI law but lack the proposed statutory underpinning of that legislation.

Why is this important?

A light touch approach will be welcomed by many stakeholders in the AI ecosystem. However, the UK does not operate in a vacuum and AI technologies are often expected to operate globally and comply with the legal obligations of major markets such as the EU and US. Where technology relies on dataset-trained algorithms, it could be all too easy inadvertently to build-in problems that become difficult to unpick later when faced with a different regulatory regime. A UK approach that tracks the fundamental principles of other major markets, but doesn't create an additional set of statutory controls, is a pragmatic recognition of global reality.

What about product safety?

While safety is flagged as a key cross-sectoral principle, the UK Government's product safety regulator, OPSS, is not listed in the proposal alongside other regulators with responsibility for designing and implementing regulatory responses. However, we know that AI is very much on the agenda of OPSS, which published a report in May this year (covered by Productwise here) addressing product safety and liability issues. This followed a call for evidence on updating the legislative framework to reflect new technologies (see here). The OPSS is expected to set out its approach to the product safety and product liability framework later this year, and it will be interesting to see whether they also pursue the "light touch" approach. Their absence from this proposal may be a sign that they are planning to chart a different course.

What next?

The UK Government is currently seeking views from stakeholders on their proposed approach to regulating AI. If you'd like to submit views, the window is now open until 26 September 2022. You may send your views to evidence@officeforai.gov.uk.

Further, there will be a White Paper and public consultation in late 2022 providing further detail of the UK Government's plans.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.