United Kingdom

[UK] Government announces initiatives to assist regulation of AI and the House of Lords introduce the Artificial Intelligence (Regulation) Bill

IRIS 2024-2:1/4

Julian Wilkins

Wordley Partnership and Q Chambers

On 6 February 2024, the UK Government published its consultation response to the AI Regulation White Paper detailing initiatives supporting individual regulators to provide tools and develop skills to address the risks and opportunities of AI. Key regulators are required to publish their plans about AI risks and opportunities by 30 April 2024. Many regulators have already published proposals; the Information Commissioner’s Office, for example, has updated data protection laws applying to AI systems. Ofcom and the Competition and Markets Authority therefore have until the end of April to publish plans to manage the technology, including AI-related risks, identify current expertise, and publish plans for regulating AI over the year. The government’s approach gives regulators autonomy to respond rapidly to emerging risks but also to enable developers to innovate. The government measures include financial support, including GBP 10 million to upskill regulators to address AI. The funding helps regulators develop research and examination tools to tackle risks and opportunities in relevant sectors like media and telecoms.

Given the pace of technological development, the UK Government wishes to avoid premature legislation implementing "quick-fix" rules that risk becoming outdated or ineffective, preferring to allow sector regulators to adapt their regulations to address AI risks in a targeted way.

The sum of GBP 90 million has been allocated for nine new UK research hubs and a US partnership to develop responsible AI. The hubs will support British AI expertise to harness the technology in different industries. The UK has committed an investment of GBP 9 million through the government’s International Science Partnerships Fund, bringing together researchers and innovators in the UK and the United States to focus on developing safe, responsible, and trustworthy AI.

Funding to the tune of GBP 2 million from the Arts and Humanities Research Council (AHRC) will support new research projects that will help to define what responsible AI looks like across sectors such as education, policing and the creative industries. These projects are part of the AHRC’s Bridging Responsible AI Divides (BRAID) programme.

GBP 19 million will support 21 projects to develop innovative, trusted and responsible AI and machine learning solutions to accelerate the deployment of these technologies and drive productivity.

A further GBP 100 million will fund the world’s first AI Safety Institute to evaluate the risks of new AI models. This builds upon the International Scientific Report on Advanced AI Safety, unveiled at the UK November 2023 AI summit, which will also help to build a shared evidence-based understanding of AI’s development and potential.

The government’s response suggests targeted binding requirements on the small number of organisations that are currently developing highly capable general-purpose AI systems, to ensure that they are accountable for making these technologies sufficiently safe.

Meanwhile, in the House of Lords, a Private Members’ Bill has been introduced entitled Artificial Intelligence (Regulation) Bill (the Bill). The purpose of the Bill is to regulate AI through a statutory AI Authority whose function would include ensuring a consistent approach amongst regulators, coordinating relevant legislation such as privacy, consumer protection and safety laws. Other functions include the appointment of independent AI auditors and supporting testbed or sandbox initiatives to help AI innovators to launch new technologies.

Under the Bill, the AI Authority would regulate AI applications to ensure a number of qualitative criteria are met including safety, security, robustness, fairness, accountability and governance. Furthermore, any business that develops, deploys or uses AI should ensure thorough and transparent testing, as well as legal compliance, including in relation to data protection, privacy and intellectual property. The Bill would regulate against discrimination whilst AI applications would have to be inclusive by design.

Within the Bill’s framework it would allow a relevant regulator to construct regulatory sandboxes for AI whereby an innovative proposition was tested in a real market situation provided there were identifiable consumer protection safeguards.

The Bill’s provisions include transparency whereby the person involved in training AI must supply to the AI Authority a record of all third-party data and intellectual property used in that training. The public must be aware whether a product or service involves AI so the consumer can give or withhold consent before use or purchase. The AI Authority must ensure long term that they engage with the public about the risks and opportunities associated with AI, and promote interoperability with international regulatory frameworks.

The Bill had its first reading on 22 November 2023 and now awaits its second reading in the House of Lords.