Me, myself and AI: Why going its own way on AI governance is an opportunity for the UK

As artificial intelligence (AI) reshapes the global economy, we are seeing intensifying competition for AI investment, talent, and capability. This competition is still in its infancy – and countries across the world are still in the process of setting out their stalls on governance and regulation. But already, we are seeing some broad contours that are shaping the global race for AI. As this race heats up, the UK has a chance to be nimble and imaginative on AI governance. Doing so could bring significant dividends.

The European Union has sought to establish itself as a regulatory leader on AI in the form of the AI Act, which was the first comprehensive legal framework around AI. This has developed a risk-based approach, under which ‘high risk’ general purpose models – those that fall under EU product safety regulation or areas that would have to be registered in an EU database – must be assessed before being put to market and throughout their lifecycle. ‘Unacceptable risk’ systems will be banned outright. AI products in general will also be subject to transparency requirements, such as disclosing summaries of copyrighted data used for training.

In contrast, the United States has embraced a more hands-off approach, prioritizing rapid innovation and deployment. Unlike the EU, there are no federal laws that directly govern AI. And early indications suggest that the new Trump administration will focus on global competitiveness through a more deregulatory approach. Indeed, Trump has already repealed Biden’s Executive Order on Safe, Secure, and Trustworthy Development and Use of AI, a pledge that was mentioned in his manifesto and statements going into November’s election.

Against this backdrop, the UK is attempting to carve out a ‘third way’ in AI governance – something between the EU’s risk-averse approach and the US’ more accelerationist philosophy. This much is clear from the references in the recently published AI Opportunities Action Plan to the UK’s pro-innovation regulatory approach as “a source of strength relative to other more regulated jurisdictions” and the need to reform the UK’s data mining framework to be “at least as competitive as the EU”. In more concrete policy terms, the Action Plan combines entrepreneurial measures like AI Growth Zones and actions to attract AI talent with prudent safety and copyright considerations, seeking to thread the needle between innovation and responsibility.

The potential prize for getting this model right is significant. By positioning itself as an ‘international bridge’ between the EU and US philosophies, the UK could establish itself as a thoughtful leader in AI governance. This can bring much more than just prestige. Despite the UK’s smaller market size compared to the EU or US, a well-calibrated regulatory environment could attract companies and investors who find the European approach too restrictive but desire more regulatory clarity than the US currently provides.

UK consumers might also gain earlier access to advanced AI capabilities – a meaningful benefit given the precedent of EU consumers facing delayed access to cutting-edge models due to regulatory hurdles. Moreover, the UK’s approach allows for tailored support to sectors where it already holds competitive advantages, such as healthcare and pharmaceuticals.

Here lies a potentially significant Brexit dividend. In pursuing these objectives, the UK now has a strategic advantage in its ability to move more nimbly than the EU – which needs to coordinate between 27 member states – and perhaps more coherently than the US, with its thus far patchwork, state-by-state approach to AI regulation.

On paper, the benefits are substantial. But several crucial questions remain as to getting the execution of this ‘third way’ right. First, the AI Action Plan has emphasised a sectoral approach to AI regulation – basically, empowering existing regulators to tackle new challenges posed by AI in areas within their scope. It remains to be seen whether this framework will provide sufficient oversight as AI capabilities continue to advance; the Action Plan concedes, “even with these initiatives, individual regulators may still lack the incentives to promote innovation at the scale of the government’s ambition”.

Second, the thorny issue of copyright presents another challenge. The government’s consultation on copyright and AI is a major early test. On one hand, major content creators are pushing hard against a copyright exemption or ‘opt-out’ system for AI data mining. On the other, an overly stringent copyright framework risks deterring AI investment or slowing model development considerably. Getting this balance right is crucial both for ensuring the UK is an attractive destination for AI and for ensuring buy-in from wider society over the long term.

Third, the UK must carefully consider just how far it wants to push on regulatory divergence. While some deviation from EU and US approaches might create competitive advantages, too much divergence could end up increasing compliance costs for companies operating across multiple jurisdictions, potentially undermining the very attractiveness the UK seeks to cultivate. Given the UK’s smaller market size compared to the EU or the US, overreach on regulatory divergence could come with costs.

The UK’s attempt to chart a middle path between innovation and regulation will of course not be an easy task. There will be many trade-offs involved to strike the right balance, and in some cases – particularly that of copyright, where the interests of two major sectors of the economy are squaring off – these will involve some big political judgement calls. But now it is outside the EU, the UK needs to make the most of its ability to provide an alternative model to Europe and America. Doing so could well be what enables the UK to continue punching above its weight on AI.