Kanishka Narayan MP, Minister for AI and Online Safety, recently stressed the importance of implementing “an AI policy which can double-down on our strengths in key areas” to make sure “we have a seat at the table” as AI capabilities develop. The Government’s focus so far has been on investing heavily in sovereign UK AI infrastructure in order to maintain geopolitical relevance as the technology advances.
In our view, the UK is also particularly well-positioned to influence global frontier AI governance conversations. As the Network Coordinator for the International Network for Advanced AI Measurement, Evaluation and Science (previously the International Network of AI Safety Institutes), the convenor of the inaugural international AI safety summit and the first country to establish a government institute dedicated to frontier AI research and model testing, the UK continues to demonstrate that international governance expertise should be valued as a core strength of its AI policy. The question now is how the UK can best double down on its leadership role to strengthen and differentiate its global influence.
One potential route to impact could be through the implementation of gold-standard frontier AI legislation. However, despite promises from two successive UK governments to institute targeted AI regulation, nothing concrete has yet materialised. The now-familiar narrative of delay surrounding the proposed frontier AI Bill reared its head yet again earlier this month when Technology Secretary Liz Kendall indicated that the Government is unlikely to bring forward new regulation beyond extending the online safety regime.
For a Government seeking to shift the narrative on UK AI to one of international ambition and confidence, we think this is an opportunity missed. As we have set out in more detail in this post, failing to regulate not only leaves the UK insufficiently equipped to address extreme risks from AI domestically, but is also a missed opportunity to cement the UK’s leadership in global AI governance.
However, an AI Bill is also far from the only route to securing the UK’s “seat at the table”. This post explores new pathways for advancing the UK’s international impact beyond simply relying on either regulatory influence or investment power, and sets out a range of options for how this agenda can be taken forward without the need for legislative intervention.
The UK’s “AI soft power”
The foundation of the UK’s outsized potential to influence global AI governance conversations relies on three strategic advantages, which together grant the UK significant AI “soft power” in its international relations:
- Strong convening capability: The UK’s global reach enables it to demonstrate notable convening power on the international stage. The 2023 inaugural AI Safety Summit brought representatives from the two great powers of AI – the US and China – together for the first time to discuss AI safety and security research. The resulting declaration committing to closer international cooperation on advanced AI has led to the establishment of the international AI safety report and the international AISI network, which the UK now leads as “Network Coordinator”.
- Close UK-US relations: The UK has succeeded in maintaining a strong relationship with the US despite changes in administration in both countries. The technology-focused Memorandum of Understanding signed in September with the US recognised that the two countries are “most trusted security and defense partners” and committed to advancing “a shared mission to promote secure AI innovation”, including through joint testing and standards development through the AISI – CAISI partnership.
- Best-in-class state AI technical capabilities: In little over two years since its establishment, the UK’s AI Security Institute has become a world-leading example of what public sector capabilities in frontier AI safety and security research and model testing can look like. AISI’s stable funding, research credentials and technical talent have allowed it to develop trusted voluntary partnerships with AI developers, which in turn compounds the UK’s international credibility and ability to create influence on the international AI stage.
Pathways for advancing the UK’s international impact
Looking beyond the usual avenues of frontier AI investment and regulation, we see three particularly promising (but not exhaustive) new pathways through which the UK can cement its position as a leading voice in global AI governance. For each pathway, we suggest examples of “low-hanging fruit”, which are ideas that the UK could implement now to advance this agenda, and “higher-ambition interventions”, which are suggestions for where the Government should begin to direct its thinking in order to maximise its long-term impact on the international stage.
Pathway 1: Strengthening AISI’s international engagement
The UK AISI has set the gold standard globally of governmental AI research and testing institutes. In its next phase, AISI should further systematise ways to diffuse its knowledge internationally. The recent announcement that the UK is stepping into the role of “Network Coordinator” of the newly renamed “International Network for Advanced AI Measurement Evaluation and Science” – formerly the International AISI Network – presents an opportunity for the UK to lead the reinvigoration of this international coordination mechanism.
- Low-hanging fruit: leverage the UK’s “Network Coordinator” role to establish strong communications channels and regular meeting structures for the network, in order to facilitate closer collaboration and information-sharing. The UK should also use its leadership position as a platform to raise the salience of emerging extreme AI risks identified by the network in other international political fora, acting as an early warning mechanism to keep governments informed of evolving AI risks and potential mitigations.
- Higher-ambition intervention: institute a separate, private information-sharing partnership between AISI and the US CAISI, building on the aims of the Memorandum of Understanding. This would act as an early warning mechanism for sharing national security-relevant risks that emerge from testing conducted by UK AISI on US AI models, to increase joint strategic awareness of emerging AI model capabilities and facilitate appropriate US responses.
Pathway 2: Advancing international alignment on risks and mitigations
The UK should continue to leverage its international convening power and relationship with the US to build mutual understanding on the evolving frontier AI risk landscape and coordinate mitigating actions where necessary. This could include proposing and building support for baseline international agreements on key frontier AI issues, for instance defining red lines for governments’ use of frontier AI capabilities, at international fora such as the UN, NATO and Five Eyes.
- Low-hanging fruit: champion best-practice regulation and research in other jurisdictions, such as the Safety & Security chapter of the EU AI Act’s GPAI Code of Practice and the Singapore Consensus on Global AI Safety Research Priorities, to inform forthcoming US AI policy interventions and facilitate high-level multilateral alignment on a baseline set of mitigations against the most prominent security risks from frontier AI.
- Higher-ambition intervention: work with AISI’s research and strategic awareness teams to identify specific extreme risks that will require closer international coordination via the establishment of dedicated global governance institutions, and secure international buy-in from key partners such as the US to create new cooperation mechanisms as necessary. One example could be a new institution dedicated to international alignment on serious AI incident monitoring and reporting frameworks.
Pathway 3: Influencing frontier AI companies’ behaviour
The technical expertise of AISI and the UK’s convening power combined make it comparatively well-placed to work with AI developers to align incentives towards effective frontier AI risk mitigation. The UK should use these relationships to strengthen its industry partnerships further and push for new voluntary commitments to raise the floor on what constitutes “best practice” for increasing the safety and security of AI systems.
- Low-hanging fruit: build on AISI’s trusted relationships with frontier AI companies to incentivise greater transparency and information-sharing proportionate to the speed at which frontier model capabilities are developing. The UK should coordinate with the companies privately to secure buy-in for extending the scope of voluntary international commitments on frontier AI risk transparency and management.
- Higher-ambition intervention: identify areas of extreme AI risk that are not adequately covered by existing regulatory regimes globally – such as the risks stemming from the race between frontier AI companies to achieve significant recursive self-improvement capabilities – and design a voluntary code of practice to help influence responsible industry practices.
This post represents the start of our thinking on this topic, not the end. As we take this work forward, we’re keen to hear from anyone working in or thinking about similar issues to help us drill down into these recommendations and find the most impactful UK policy interventions for advancing global AI governance.
