UK’s global AI safety body hopes dented by summit communiqué

Prime Minister Rishi Sunak’s ambitions to establish a new global AI research body appear to have been dealt a setback, according to a draft communiqué obtained by POLITICO before next week’s flagship AI Safety Summit.

According to an Oct. 26 Politico report, the document — dated October 25 and presented to EU member states — acknowledges the potential for “serious, even catastrophic” harm from advanced AI systems. However, it signals that further study could be carried out through existing multilateral efforts like the United Nations and the Global Partnership on AI rather than a separate entity.

This represents a change from an earlier draft, which only referenced the network “supporting” collaboration and did not mention the UN. The UN announced its own high-level advisory body on AI in August, due to report back in 2024.

Holding future AI safety summits seems to still be in question

The latest version of the Politico report also references holding future AI safety summits in square brackets, suggesting a lack of agreement among participating nations. Earlier this week, UK Technology Secretary Michelle Donelan said she hoped the November 1-2 summit would be the “first of many.”

Other tweaks appear geared towards aligning with positions held by China and the EU. Mentions of UNESCO and the G20 have been removed, while language around risk-based governance frameworks has been added.

An EU official stated these edits reflect a desire to adhere closely to the bloc’s AI Act legislation. The Act takes a proportionate, risk-based approach to regulating AI based on risk categories.

The changes are a potential setback for the UK’s ambitions to spearhead global coordination on AI safety via a dedicated research body. However, a UK source disputed this interpretation, claiming the new wording supported their goals.

We watch for the final tally, which will be out next week. We look forward to reporting the results here.

The final verdict will become evident when the summit communiqué is published next week. However, the revised draft suggests an uphill battle for the UK in achieving international consensus on addressing potential risks from advanced AI.

Featured Image Credit: Monstera; Pexels; Thank you!

The post UK’s global AI safety body hopes dented by summit communiqué appeared first on ReadWrite.