Written Evidence Submitted by CBI







Across the UK, the CBI speaks on behalf of 190,000 businesses of all sizes and sectors. The CBI’s corporate members together employ nearly 7 million people, about one third of private sector employees. With offices in the UK as well as representation in Washington, Brussels, Beijing and Delhi, the CBI communicates the British businesses voice around the world.


The UK’s economy is facing stark challenges from labour shortages and supply chain freezes to the fallout of a war in Europe and high energy costs. The CBI’s latest forecast suggests that GDP will contract by 0.4% this year1. With the Prime Minister’s 2023 pledge to grow the economy, the UK needs a laser focus on sustainable, high growth that leverages and builds UK strategic advantage.


Achieving growth will require putting innovation at the heart of the economy. Innovative sectors, like the UK’s $1tn digital economy and world-leading Artificial Intelligence (AI) research base, must be well equipped to deliver economic growth and boost productivity, including through the £38bn on offer from widespread AI diffusion across the economy2. By achieving this, the UK can also address complex societal challenges such as climate change and an overburdened healthcare system.


Smart regulation, that balances certainty with flexibility, is the key to unlocking the AI prize across the UK economy. AI is a prime example of where the UK can leverage its existing strengths to spur innovation, boost investment confidence, and strengthen UK strategic advantage in an emerging global field. Considering AI governance in its fullest sense - encompassing a wider ecosystem including the role of standards, developing an assurance ecosystem, alongside principle-based regulation - is critical towards this aim. As it sets a national approach to AI governance, the UK must make clear, strategic, future-proof decisions that are fit for dynamic, rapidly changing AI technologies. The Government’s intended approach3 has been welcomed by business across the economy, particularly in comparison to the EU AI act.










1 CBI, Economic forecast, December 2022

2 CBI, Seize The Moment, May 2021

3 HM Government, Establishing a pro-innovation approach to regulating AI, July 2022


While the direction of travel is right, UK AI governance risks running into certain pitfalls including regulatory overlap, fragmentation and complexity. As AI governance moves towards detailed principles and implementation, businesses have set out three recommendations to avoid these pitfalls and support a pro-innovation, proportionate and coordinated approach across AI use cases.


AI presents significant economic and societal benefits for the UK. The UK is at a pivotal moment and must seize the opportunity to make clear, strategic decisions about how AI will be governed

Government has set the right initial direction on AI governance but must now look towards implementation

AI has quickly developed into a mainstay of modern life which is helping to address complex societal and economic challenges. Yet current AI governance is diffuse, varied, and unclear with no national overarching regime – which firms find difficult to navigate. Some sector regulators have mature AI workstreams while others have nascent efforts leading to little cross sectoral consistency. In supporting an innovative AI sector and creating an environment where businesses expand their use of AI, the Government’s proposals for a light-touch, context-specific, outcome driven approach led by sectoral regulators is the correct choice. The approach will leave implementing specific regulation to existing sectoral regulators, where they are better placed to deal with specific AI context than a new statutory AI regulator. The Government also intends to use cross-sectoral principles rather than legislation to create a light-touch, outcome led approach with risk based, proportionate regulation as opposed to an innovation-stifling one size fits all solution.

With Government having successfully identified the right challenges and chosen the right approach, implementation will need to harmonise both aspects.


Government now needs to take the next step and turn principles into policy. Decisions taken now will have a lasting impact on AI innovation and diffusion in the UK and set the basis for further developments going forward. To avoid needing frequent reiteration or creating uncertainty for businesses it is therefore essential that the national regime embeds innovation, proportionality, and coordination.


Businesses continue to face challenges navigating emerging frameworks governing AI

Businesses understand the value of smart regulation to provide clarity and common guardrails when developing, investing in, and adopting AI. Many firms are already considering and implementing their own ethics and governance regimes to ensure their use of AI is fair, ethical and delivers good outcomes for customers and clients4. Some firms have created new AI ethics roles and teams, highlighting the importance of implementing these values into the specific AI use cases of individual firms.




4 CBI, AI: Ethics into practice, August 2019


At a national level however an approach determined exclusively by sector regulators would not address certain challenges that businesses face including regulatory overlap, fragmentation, and regulatory complexity. These challenges are likely to be particularly acute in uses cases that sit across multiple sectors or vendors that sell to multiple sectors, such as cloud computing providers. As digital regulation develops, businesses are feeling the burden of navigating an increasingly complex landscape including determining where they are in scope and what their compliance responsibilities are. Firms developing and adopting AI are looking at a range of approaches from regulators and policymakers. A financial services firm for example could see the ICO AI Auditing Framework, upcoming changes to GDPR on automated decision making processes, and the

FCA’s ongoing development of AI guidance as all relevant to their work. In this circumstance there is no clear direction for which set of guidance or regulator takes precedence and where businesses should go to first. This creates regulatory complexity and uncertainty which stifles both innovation and investment decisions. Government and regulators will need to work together to ensure AI governance does not increase overlap and fragmentation by mapping out existing guidance and helping firms to navigate the ecosystem.


An AI Governance Hub would strike the right balance between decentralisation and coordination

The complexity of the regulatory landscape can make it challenging for businesses of all sizes to understand the objectives of their regulator and how best to approach AI governance. It is vital that AI governance plans and objectives are well-communicated to regulators and integrated into their workplans. This will increase regulator ability to work with the businesses they regulate to minimise confusion and reduce the barriers to entry for new innovative businesses and products.


To avoid these shortcomings, it is important that a central framework provides resources for both regulators and businesses while coordinating activity across the AI domain. The central framework should aim to become the one-stop shop for the AI governance needs of businesses, regulators, and consumers. The central framework should not have enforcement powers or a formal statutory footing, in order to retain the benefits of decentralised regulation, but could take many forms. One form the Government should consider is an AI Governance Hub (AIGH) which:























5 CBI, The Office for Future Regulation, 2022


A successful UK AI governance regime will need to blend principles and sector specific guidance

The nature of AI creates unique difficulties in its governance

As a general purpose and quickly evolving technology, AI has the potential to provide solutions to both specific societal challenges, such as reaching net zero, and UK economy wide problems, including boosting productivity. This flexibility is a strength but also creates unique difficulties for AI governance. Concrete regulations will quickly become out of date and are inappropriate for a sector-specific, outcome driven approach with a non-statutory approach thus favoured. A lack of shared core narrative for AI governance however risks creating misalignment between regulators and not addressing the problems with the current system. Balancing these concerns will involve a careful blend of principles and guidance.


Cross sectoral principles address concerns associated with having no horizontal regulator

Without a horizontal regulator the most appropriate way to guide work in different sectors towards similar aims is through cross-sectoral principles. The Government’s currently proposed principles are: ensure that AI is used safely, ensure that AI is technically secure and functions as designed, make sure that AI is appropriately transparent and explainable, embed considerations of fairness into AI, define legal persons’ responsibility for AI governance, and clarify routes to redress or contestability.


Whilst firms welcome these principles and support their alignment with the OECD value-based principles6, there is notably no principle recognising the potential for AI to boost innovation and create sustainable growth in our economy. A growth-related principle should be added to reflect the overarching aims of the Government – including leveraging AI as a priority focus for the Department for Science, Innovation and Technology, and the UK AI Strategy. The OECD principle of Inclusive Growth, Sustainable Development and Well-being could be used as a guide. The current absence risks overlooking the potential for AI to address complex societal challenges and support growth in the UK, leading to an unbalanced regulatory approach to guidance and enforcement. Inclusion would thus recognise the risk of not using AI to address UK challenges and provide a framework for more balanced outcomes.


Specific language within these cross-sectoral principles should be carefully considered. Businesses support Government’s proposal which notes that these principles are not intended to create an extensive new framework of rights for individuals, however the Government’s draft principles include wording that has contextual legal grounding, for example ‘fairness’ as a legal concept when applied to personal data. It will be necessary to monitor these principles and issue guidance where necessary. This should be a part of ongoing efforts to develop an outcomes monitoring framework for digital regulation7. The same principle can also mean different things to different audiences, for example, the general public and AI specialists will have different standards for what is explainable. Principles need to be flexible enough to reflect these possibilities.


Central guidance provides a framework to create certainty for regulators and businesses

Under the proposed AI governance system most specific implementation will sit with sector specific regulators and the AIGH should set clear boundaries of remits to prevent regulatory creep. Where guidance from the AIGH will be useful is helping relevant regulators to apply the cross-sectoral principles of AI governance into their own sector. Specific guidance would give clear direction to regulators and businesses, overcoming some of the barriers of a non-statutory approach. The


6 OECD, OECD AI Principles Overview, 2019

7 DCMS, Plan for Digital Regulation: Developing an Outcomes Monitoring Framework, June 2022


AIGH should look to the Alan Turing Institute’s AI Standards Hub8 as a positive model of how, with the appropriate structure and oversight, a similar hub can become useful for firms to share, amend, and adopt the latest pro-innovation compliance methods. Certain areas where guidance should be considered includes:





Businesses need a way to voice ongoing concerns and raise queries to support better governance outcomes

Regulators and government should seek to increase the level of structured and direct engagement with business and civil society. The nature of AI and difficulty of implementing governance processes suggest teething issues are likely to occur. Business engagement creates a direct link between those making the regulations and those that are applying them, and ultimately helps deliver better compliance outcomes. Engagement during both the development and implementation processes would lessen these issues. Routes to more direct engagement could include a business forum operated by the AIGH. The forum would act as a focal point to discuss cross-cutting issues with multiple regulators and to voice concerns that otherwise lack an obvious relevant regulator. This approach would unite relevant stakeholders to improve coordination while enabling industry leaders to address challenges of implementation.


8 AI Standards Hub, About the AI Standards Hub

9 National Institute of Standards and Technology, AI Risk Management Framework, 2023


Becoming a global leader in AI requires international coordination and engagement

UK consumers and businesses need access to the best AI systems

AI systems are quickly evolving around the world, and it is to the benefit of UK consumers and businesses to have access to the best AI systems regardless of where they are developed.

Interoperability of AI regulation is the best way to achieve this however differing approaches to AI governance however are quickly forming, with the EU AI Act and US AI Bill of Rights key examples. With potentially conflicting regulation and significant compliance burdens there is a risk that parallel AI frameworks emerge. This scenario could:


Standards are an evolving way to build trust and interoperability

Voluntary industry-driven technical standards provide a non-regulatory pathway to give firms clarity, certainty, and consistency in using and developing AI. Standards establish best practice and support harmonisation. Standards also avoid burdensome regulatory approval processes and can be globally orientated to avert the need for complicated adequacy systems, as currently occurs within data protection. The development of the infrastructure to support AI standards, including the UK Government supported AI Standards Hub, is well underway and will be a critical resource to coordinate UK involvement in standards-setting as they are developed, as well as business use of international standards. Many standards are still under development and the standards available will take time to mature. Accordingly, it would be inappropriate for standards to be made mandatory which could freeze develop and reduce the flexibility advantage that standards provide.


Developing an overarching AI assurance system is an ongoing but necessary step towards become a world leader in AI

Development of AI standards should continue to include the wider AI assurance ecosystem. AI assurance services are the tools, services and professionals needed to build a trustworthy AI ecosystem such as auditing and certification. Establishing this ecosystem is critical to build trust amongst the consumers, businesses, and wider public who will be adopting AI in the coming years. Different branches of the assurance ecosystem will manage aspects of AI risks not fully covered by standards or the AIGH including developing an AI assurance market. This will build on the strengths of the UK’s existing professional and technology sectors and play an important role in avoiding hard line, statutory regulation. The Centre for Data Ethics and Innovation’s Roadmap to an effective AI assurance ecosystem10 has set out a strong path forward with notable progress across its strands of work. As it continues to propel the UK towards more developed AI assurance, CDEI must double down on industry engagement. Input from firms will be vital to make sure the UK not only has the most trustworthy ecosystem but also becomes the best regime to develop and adopt AI in.











10 CDEI, The roadmap to an effective AI assurance ecosystem, December 2021


At this early stage, the UK is well positioned to shape the discussion on the future of AI governance and should seize the opportunity to become a world leader. AI is globally orientated by its nature and the UK should use its position on international fora like the G7 to deploy its AI governance expertise to influence international trends on AI regulation and interoperability. This intention must be backed up with continued support for the AI assurance ecosystem to extend a UK strategic advantage and carve a clear path for others to follow. With the AI assurance ecosystem in place and a well-designed AI governance system implemented, the UK will be fast tracking the National AI Strategy’s aim to ensure the UK gets the national and international governance of AI technologies right to encourage innovation, investment, and protect the public and our fundamental values11.




(February 2023)