The UK government (the "Government") has published proposals for a new regulatory framework for artificial intelligence (the "White Paper"). Its goal is to "provide a clear, pro-innovation regulatory environment" to make Great Britain one of the "top places in the world to build foundational AI companies". The White Paper outlines a common-sense approach aiming at fostering innovation without compromising safety or privacy.
Concerns about risks posed by artificial intelligence (AI) have been raised by regulators and industry. Notably, an open letter endorsed by hundreds of influential figures has discouraged the development of "human competitive" AI systems, citing fears of harmful disinformation, jobs automation and human obsolescence. Italy's data protection regulator has temporarily banned ChatGPT, citing data protection concerns, and the UK's Information Commissioner issued fines against US firm Clearview for its AI-powered facial-recognition software.
The European Commission has proposed its own draft Regulation harmonising AI regulation across the bloc (the "AI Regulation") which will broadly prohibit AI applications with an "unacceptable risk" level. This article considers the White Paper proposals and highlights the differences from the AI Regulation.
The Proposed UK Regime
The UK's regulatory framework will rest on five key principles:
- Safety, security and robustness – AI systems should function in a robust, secure and safe way throughout a life cycle.
- Appropriate transparency and "explainability" – AI systems should be appropriately well explained and sufficiently transparent.
- Fairness – AI systems should not undermine the legal rights of individuals or organisations or discriminate unfairly against individuals or create unfair market outcomes.
- Accountability and governance – AI systems should be effectively overseen and clear lines of accountability established.
- Contestability and redress – Parties should be able to contest an AI decision which risks harmful outcomes.
The key principles will be non-statutory at first, and regulators are only advised to consult them when determining AI regulation. But regulators may become duty-bound to consider the key principles after an initial implementation period subject to Government review.
Comparing the UK and EU
The Government proposes a "common-sense, outcomes-oriented approach" balancing the aim of making the UK an AI superpower by 2030 against "proportionately regulating" serious risks posed by AI. The UK is taking a markedly different approach than the EU:
- Broader definition. The White Paper defines AI as any system with functional capabilities that combine "adaptivity" and "autonomy". The broad definition ensures the regulations can be applied—and amended—agilely to evolving technologies. The AI Regulation instead uses a fixed and narrower definition of AI.
- Context-specific. The White Paper avoids product-specific categorisation, instead preferring outcome-based regulation. Applications of AI will be contextualised and considered as either high or low risk on a case-by-case basis, and the regulators will use this to balance the benefits against the risks. In contrast, the AI Regulation proposes a formal taxonomy of risk (from low to high) and bans the use of AI technologies deemed to pose an "unacceptable" risk.
- No new regulators. The decentralised UK regime will not create a new regulator; instead, it will rely on existing regulators to enforce the Financial Conduct Authority, the Information Commissioner's Office, the Competition & Markets Authority, the Equality & Human Rights Commission and the Medicines and Healthcare Products Regulatory Agency. Contrastingly, the EU proposes both a central regulator (the European Artificial Intelligence Board) along with national supervisors in each member state.
- No new penalties. The White Paper proposes no new penalties for infractions. Instead, the various UK regulators will use their current enforcement powers. By contrast, the EU regime introduces new penalties of up to 6% of global annual turnover or €30 million (whichever is highest).
- The Government's role. The Government will provide central support functions to UK regulators including monitoring for effectiveness and coherence, assessing economic impact and promoting interoperability with foreign regimes. The White Paper proposes that the Government will collaborate with the regulators in implementing its proposals. Its role will include identifying the risks of AI application, working with regulators to identify cross-sector "central risks" and updating laws and regulations to respond to risks flagged by regulators.
The White Paper raises several issues:
- Will the existing regulators have the capacity to take on the additional responsibilities?
- Will the existing regulators have the expertise to review the compliance measures of technically complex AI applications?
- Will the alphabet soup of regulators create inconsistencies in the application of the key principles as well as overlaps or lacunae in regulation?
The UK is regarded as one of the most "AI-ready" nations, and its proposed principles and outcomes-based approach, underpinned by behavioural economics, have been successful in other contexts, such as in financial services. The Government has adopted a similar approach to digital regulation more broadly, notably in respect of digital platforms, data privacy and online safety1. The Digital Markets Unit within the Competition & Markets Authority will be underpinned by statute and formally launched over the coming months2.
As both the UK and EU become more active in AI regulation, the two divergent systems may create incoherence as well as additional compliance burdens for businesses operating in the EU, in particular.
The Government has launched a public consultation of the White Paper's proposals, inviting stakeholders and the public to respond to the consultation questions by the deadline of 21 June 2023.
With thanks to trainee Ben Hasson who also contributed to this article.
2. The Digital Markets, Competition & Consumer Bill is due to be under legislative scrutiny this year.
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.