UK AI Developments: Preparing for the AI Safety Summit

Safety First: Frontier AI

Why should I read this?

The UK is preparing to host its Global AI Safety Summit 2023 on 1 and 2 November at Bletchley Park, Buckinghamshire  (the Summit) bringing together international governments, leading AI companies, civil society groups and experts. The focus of the Summit will be on how to best manage the risks of recent advances in AI, specifically frontier AI.

Following its “pro-innovation  approach to AI regulation” whitepaper published in March 2023, the UK Government hopes that the Summit will shape global AI discourse. The whitepaper sets out the UK government's general approach to regulating AI domestically. It focuses on a principle-based regulator-led governance of AI with a sectoral focus and only limited central government involvement nor any general AI laws (further discussed: here). The whitepaper acknowledges the cross-border nature of the subject matter and sets strengthening the UK’s position as a global leader in AI as one of three main objectives.

In this article, we break down the key concepts of the UK’s approach to the Summit.

What should I do?

How AI is to be governed, monitored and regulated globally continues to be an ongoing debate. The Summit presents another step in this debate. It will be important to keep the Summits’ objectives and the risks being considered in mind as national and international frameworks, guidance and regulations develop.

The Summit focuses on frontier AI - highly capable general-purpose AI models that can perform a wide variety of tasks and match or exceed the capabilities present in today’s most advanced AI. Current examples of frontier AI include the foundation models that form the basis of generative AI systems such as ChatGPT. Additionally, the summit will be looking at frontier AI in certain specific high-risk “Narrow AI” use cases  (e.g. AI models for bioengineering). 

The UK Government’s objectives for the Summit include: 

1. achieving a shared understanding of the risks posed by frontier AI
2. creating a process for international collaboration 
3. outlining measures for organizations to take on frontier AI safety
4. identifying areas for collaboration on AI safety research, including on development of governance standards
5. showcasing how safe development of AI can be used for good globally

Prime Minister Rishi Sunak noted in his speech on AI (26 October 2023) that the UK will aim to agree upon an international statement about the nature of AI risks. He underscored the UK’s ambition to establish a global Intergovernmental Panel on AI (similar to the UN-led Intergovernmental Panel on Climate Change). This Panel would be tasked with publishing a State of AI Science report, to reach an international scientific consensus on AI   developments, potential impacts and response options. It will be interesting to watch at the Summit how this proposal works together with (or as part of) the UN’s announcement of an AI Advisory Body.

What are the key risks being considered at the Summit?

In anticipation of the Summit, the UK Government has also published a discussion paper on the capabilities and risks from frontier AI including an assessment by the UK intelligence communities. Its report identifies four categories of possible risks and harms that will be discussed at the Summit: 

1. cross-cutting risks across sectors, including the lack of AI safety standards in the design and implementation of AI
2. societal harms, such as bias and fairness
3. the misuse of frontier AI, with respect to for example dual-use technology and cyber
4. loss of human oversight and control through misalignment or intentional subversion by an AI system

What else do I need to know about the AI Safety Summit and how will it impact UK regulation?

While the primary focus of the Summit is on international cooperation, it is expected to serve as a platform for the UK Government to articulate its position and share insights on its ongoing AI regulatory strategy. 

The Summit will be a platform for discussions on frontier AI with stakeholders including technical experts, governments and the private sector. While the Summit is not focused on regulation, it is likely to shed light on the Government's stance and plans in regulating AI, setting the stage for future regulatory developments both on the international stage and in the UK. 

If you have any questions relating to this or AI more generally, please get in touch with any of the contacts listed below.

Thank you to Felix Schulte-Strathaus for helping to draft this update.

Further reading on UK AI regulation:

• AI UK white paper and regulatory guidance – the ICO response
• UK government’s white paper on AI – innovation balanced with protection from harm
• UK Competition Authority launches review into AI