A draft agenda of the impending AI Safety Summit next month has revealed the potential intention of establishing an ‘AI Safety Institute” as part of its international collaboration.

Advisors to the Prime Minister have been attempting to lock down international collaboration for the future of AI regulation which will be the main topic of discussion at the AI summit soon to be held at Bletchley Park this November.

Specifically, the Summit is set to discuss the associated risks of AI and its various forms, and discuss how the advancements of AI could affect national security.

The “AI Safety Institute” would focus on this last point, specifically how advancing frontier models could be regulated by nations in the interest of international security and collaboration.

Part of the ambitions announced for the summit include this topic, with the intention of creating a forward process for international collaboration on frontier AI safety, including how best to support national and international frameworks.

Matt Clifford, the Prime Minister’s representative for the AI Safety Summit, however, somewhat countered the idea of setting up a formal institution as suggested by the draft agenda which was seen by The Guardian, saying in a post on X that “Our view is that most countries will want to develop their own capabilities in this space, particularly to evaluate frontier models.”

So far, the Summit is expected to host around 100 people, ranging from AI professionals to cabinet members from various governments.

Companies expected to participate in the summit include ChatGPT developer OpenAI, Google and Microsoft and it is said they will publish details on AI safety commitments, agreed with the White House in July, with updates likely to include reference to safety, cybersecurity and how AI systems could be used for national security purposes.

A government spokesperson said: “We have been very clear that these discussions will involve exploring areas for potential collaboration on AI safety research, including evaluation and standards.

“International discussions on this work are already under way and are making good progress, including discussing how we can collaborate across countries and firms and with technical experts to evaluate frontier models. There are many different ways to do this and we look forward to convening this conversation in November at the summit.”

Recommended reading

• Tech Professionals: PM Should Make Ethics a Priority at AI Safety Summit
• UK’s AI Safety Summit Ambitions Announced
• UK Government Adds Industry Leaders for Inaugural AI Safety Summit

While it is still uncertain what the summit will result in – either further regulation or international bodies – collaboration is key, according to Clifford.

“Collaboration is key to ensuring we can manage risks from Frontier AI – with civil society, academics, technical experts and other countries,” he wrote on X.