AI regulation: where are we at & what’s next?

The UK is at the forefront of AI development and research, ranking third in the world after the USA and China. According to the International Trade Administration, our AI industry is projected to contribute $1 trillion to the UK economy by 2035 – a staggering number that speaks to the exponential growth rate of the sector.  

It’s no surprise then, that the government has put AI at the front and centre of its agenda. In 2021 it published its first 10-year National AI Strategy, which set out plans to cement the UK as an ‘AI superpower’ through investment and a digital-first approach. Since then, we’ve seen a flurry of government activity in this space, including the appointment of the first Minister for AI and IP.  

But at the heart of this activity lies one key question – how can governments foster growth and innovation, while also protecting consumers and creating a framework for safe application of AI technologies?  

AI regulation in the UK 

As it stands, the use of AI in the UK is regulated in the context of its application, through existing legal frameworks like financial services regulation. But these frameworks weren’t designed to address the new and increasingly pressing questions that the application of AI creates. Regulators are grappling with thorny questions such as who owns the rights to music generated by AI? Who is responsible for the harms created by algorithmic bias? The list goes on.

The Government’s AI White Paper, published in March 2023, set out how it plans to bring our regulatory approach up to date. In order to strike a balance between rigid regulation and innovation, the White Paper establishes five values-focused regulatory principles, including fairness, accountability and security. Existing regulators will be required to apply these principles proportionately to risks posed by AI within their respective remits and in accordance with existing laws and regulations. We’re already seeing regulators implementing this new approach – the Information Commissioner’s Office (ICO) in particular has been quick out of the blocks, confirming that AI is one of its top priorities and publishing extensive guidance and resources for businesses.

This principles-led approach, which the Government describes as ‘regulating to innovate’, is designed to deliver sector-specific solutions that can adapt to developing applications of AI and align with wider international frameworks.  The focus on creating a future-proofed, flexible framework is welcome and feels like the right next step as it should allow regulators the freedom to take an approach that best suits the needs of their individual sectors.

That said, this approach creates a risk that the regulatory landscape and approach to enforcement across different sectors could become inconsistent and confusing, and that needs to be carefully managed. The success of the Government’s approach will therefore be dependent on close working between government and regulators, including through the Digital Regulation Cooperation Forum, to ensure that there is parity in the application of the new principles.

But we must also acknowledge that the tech industry has a critical role to play. Regulatory principles can take us so far, but the first global guidelines for the secure development of AI technology, published in November, emphasise the importance of AI developers building programmes that are cyber-secure from inception, or ‘secure by design’. In other words, a successful future regulatory framework for AI will need to be tripartite, with government, regulators and industry all playing their role in making AI safe.

Change on the horizon

In February 2024 the Government set out some next steps for AI regulation in a response to its AI White Paper consultation. The response reiterates the Government’s commitment to a principles-led approach and focuses on implementation, with commitments related to monitoring (including consulting on a new risk register) and support for regulators.  

It’s clear that a non-statutory approach remains the core of Government policy, but there are hints that we may see divergence from this in future. Highly-capable general purpose AI systems are identified as a real regulatory challenge, and the Government acknowledges that there is likely to be a need to consider risk thresholds and binding requirements on the developers of this technology in the future.  

Meanwhile, Labour has been vocal about the huge economic opportunities of AI technologies, but has also been cautious about the risks, particularly around large language models. In a speech made last year, the then Shadow Minister for the creative industries, Lucy Powell, indicated that Labour would back a licensing regime for large language model development that would be overseen by a new, dedicated regulator. Around the same time, Labour leader Keir Starmer set out his intention to take a tougher stance on AI, and trailed plans for an ‘overarching regulatory framework’.  

While there’s no suggestion that we’re heading for an EU or US style regulatory overhaul, it’s clear that the UK is only just setting off on the journey to finding the right regulatory ‘fit’ to create a sustainable and safe environment for AI application. Where that journey will end is unclear – even ChatGPT can’t answer that one.