UK proposes ‘light touch’ approach to regulating artificial intelligence

Like an elephant, the nature of AI is hard to put into words, but you can see it when you see it.* This poses a challenge to policymakers seeking to leverage regulation to support the safe development of artificial intelligence. In a new policy document, the UK government emphasized flexibility in its approach without clarifying what it means by AI. But the EU’s initial proposals for AI-related legislation may end up setting a global standard.

Pro-innovation regulations

The Department for Digital, Culture, Media and Sport has set out an ‘innovation-friendly’ vision for the future regulation of artificial intelligence in a policy document. This paper sets out the components of a cross-sectoral regulatory framework.

As the paper points out, there is no law explicitly written to regulate AI in the UK. This means that companies rolling out AI systems must ensure they comply with existing legal and regulatory frameworks. For example, the Information Commissioner’s Office (ICO) took action against Clearview over its facial recognition technology and promised to investigate concerns about the company’s use of algorithms to screen job applications. See also the note on how AI in financial services is regulated in the UK.

However, a lack of regulation around AI can cause confusion and stifle innovation. Respondents to a 2019 survey of financial institutions suggested that additional guidance on how AI fits into existing rules could encourage more companies to adopt AI.

To ensure consistency across different industries, DCMS plans to set cross-sector principles for AI and ask regulators to contextualize these for the sectors they oversee.

Principles and guidelines, not rules

DCMS has set out six guiding principles for regulators to consider when overseeing the use of AI in the sector. these are:

1. Make sure AI is used safety
2. Make sure you have AI technically safe Features and functionality as designed
3. Make sure you have AI. moderately transparent And explainable
4. Includes considerations equity into AI
5. Defining corporate responsibilities for AI rule
6. Make your path clear salvation or potential contention

DCMS does not expect that these principles will necessarily be translated into new obligations. Instead, we plan to encourage regulators to prioritize lighter touch options, such as guidance and voluntary measures. Regulators have been told to adopt a proportional, risk-based approach focusing on high-risk issues.

This flexible approach is likely to be well-received, but choosing not to regulate this area could result in the EU’s stricter rules becoming the de facto standard for AI regulation.

EU differences

Unlike the EU, the UK is not preparing to introduce AI-related legislation. Instead, DCMS suggests that responsibility for designing and implementing a proportionate regulatory response should be delegated to regulators.

The European Commission’s bold proposals for AI law aim to regulate AI systems across the EU according to the level of risk they present. The draft bill would ban AI systems that present unacceptable risk, impose strict requirements on systems deemed high risk (such as those used to assess credit risk or provide credit scores), and It aims to apply transparency requirements to potentially low-risk systems.

The EU framework could bring about major changes, requiring companies to assess the risks of AI systems and comply with relevant obligations. Failure to meet the requirements for high-risk AI systems may result in fines of up to €30 million or 6% of global revenue, whichever is greater. Find out more about what the EU is doing to foster human-centric AI in our blog post.

Another difference between the EU and the UK is their approach to defining AI. While EU AI law contains a very broad definition, the DCMS policy document chose not to define AI. Instead, it points to key characteristics of AI technology that existing regulations may not be fully suited to address.

These characteristics are:

• adaptability That said, the logic behind the output can be difficult to explain.
• autonomy That is, the ability to make decisions without explicit intent or human intervention.

It is this combination of characteristics that calls for a tailored regulatory response to AI. By focusing on these core characteristics, DCMS argues that there is no need for a detailed, universally applicable definition of AI.

DCMS acknowledges that its proposals differ from the vision of AI regulation set out by the EU, but says the EU’s approach of setting “relatively fixed definitions” in law does not capture the full application of AI regulations and therefore leaves the UK with a different vision of AI regulation. They claim it wouldn’t be suitable. AI and regulatory implications.