FTI Consulting Public Affairs Snapshot: A successful summit for Sunak, but can Britain stay at the helm of international AI policy?

However, with fierce competition for the spotlight and a lack of domestic legislation, it is far from certain that the UK will hold on to this position in the months ahead.

A key success was the Bletchley Declaration, the first global AI safety agreement, which gained the support of 28 nations, notably the US and China. It will drive forward an international consensus on AI risks and commits to supporting an “internationally inclusive network of scientific research on frontier AI safety”. However, the level of China’s participation remains uncertain given UK and US hesitance over sensitive tech collaboration.

Separately, a “landmark” agreement was reached between governments and leading AI companies to commit to model testing. The summit’s legacy as a forum for international cooperation was also secured, with two more AI Safety Summits scheduled in South Korea and France.

Throughout the gathering at Bletchley Park, Sunak not only successfully positioned AI as a permanent feature at the top of the international political agenda, but he also placed Britain at the helm of the discussion – at least for now.

Indeed, the summit was not only about securing a leadership role for the UK, but also a legacy for the Prime Minister himself. With current polling strongly favouring Labour, Sunak has secured AI leadership as an enduring marker of his premiership, whatever happens at the next election.

The summit delivered various commitments and grand statements while putting the UK at the forefront of AI diplomacy. But the question of whether such influence will endure is more complex. Labour’s Shadow Technology Secretary, Peter Kyle MP, has pointed out, for instance, that the agreements made at Bletchley were not legally binding.

Only time will tell how strongly signatories will stick to their commitments and, crucially, if tech companies will meaningfully live up to their promises. Experience with international agreements on climate change and the annual COP conferences speaks to the fact that even legally binding accords are not always upheld.

What’s more, behind the friendly diplomacy on display last week, the global race to scale up sovereign AI capabilities rumbles on as the West and China continue to compete and decouple from one another.

While the UK has undoubtedly got in early with this summit, there is no guarantee that it will continue to set the course on AI regulation. A hawkish speech by US Vice President Kamala Harris, for instance, was a display of US muscle in this space, with Harris affirming that US domestic AI policy should “serve as a model for global policy”.

The US position, unlike the UK’s, is backed up by concrete intentions through a sweeping Executive Order, as well as its position as home to the world’s largest AI sector. Any joint initiatives, such as the UK-US AI Safety Institute partnership, will more likely be US-dominated.

Meanwhile, China’s spokesperson, representing the world’s second-largest AI sector, gave a speech about the significance of its own Global AI Governance Initiative. Even Italy, previously averse to generative AI, took the opportunity to announce an upcoming AI conference during its G7 presidency next year.

While international competition for the AI safety spotlight demonstrates the success of the summit in driving the international conversation, it also highlights the inherent contradiction between the UK Government’s rhetoric and action on AI. While the summit sought to address the existential risks of the technology, the Prime Minister has repeatedly said that the UK will not be rushed into regulating AI.

At the summit’s conception, Sunak pitched the UK as the natural bridge between the US, EU, and China. At the time, his officials cited Britain’s departure from the EU as allowing it to act quickly in response to the rapidly changing developments. Fast forward five months and it is evident that the UK is lagging behind in domestic legislation.

The EU’s Artificial Intelligence Act (AIA) looms large as the first significant piece of legislation on AI. The impact of the EU’s first-mover advantage was clear as it exercised its lobbying power in the lead-up to the summit by amending the wording of draft iterations of the Bletchley Declaration to include that proportionate governance policies could include “risk-based” frameworks. A risk-based approach was also endorsed in the G7’s code of conduct, which naturally received the EU’s support.

Furthermore, the gap between the traditionally low-regulation US and the highly-regulated EU is shrinking. President Biden’s Executive Order on AI regulation was more stringent than many expected. This increases the pressure on the UK to strengthen its proposed non-statutory principles-based approach to AI regulation when the Government responds to the AI White Paper consultation before the end of 2023.

But with approximately a year until the next election, the opportunity for Sunak’s administration to bring forward domestic legislation is slipping away. Given the Government’s comments about not rushing to regulate AI, an AI Bill is not expected to be included in tomorrow’s King’s Speech. This leaves amendments to the Data Protection and Digital Information Bill as the most likely legislative hook before the end of the next Parliamentary session.

Ultimately, behind the UK’s successful hosting of this first AI Safety Summit, an international consensus among Western nations for a risk-based and stringent approach to AI regulation emerged. Domestically, the UK will therefore face pressure to fall in line to adopt such an approach, pivoting away from its proposed non-statutory principles.

With the response to the AI White Paper anticipated before the end of the year, this will be a key test of whether the UK can indeed capitalise on the success of the summit and produce leading domestic proposals, making the next few months crucial for the development of Britain’s AI policy.