Bletchley Declaration on AI safety cooperation signed by 28 countries at UK summit

Thu 2 Nov 2023
Posted by: Danielle Keen
Trade News

After the opening day of the UK’s AI Safety Summit (1 November), 28 countries and the EU have agreed to cooperate on the responsible development of AI.

The Bletchley Declaration, now the first global statement on AI safety, acknowledges that AI can “transform and enhance human wellbeing, peace and prosperity”, while also having the “potential for serious, even catastrophic, harm”. To mitigate this risk, signatories have committed to international cooperation and information sharing as technological advances are made.

UK prime minister, Rishi Sunak, said: “This is a landmark achievement that sees the world's greatest AI powers agree on the urgency behind understanding the risks of AI – helping ensure the long-term future of our children and grandchildren.”

Signatories

The statement boasts global representation with nations from all continents agreeing to its terms. This includes African nations, such as Nigeria, Kenya and Rwanda; countries from the Middle East, including Saudi Arabia and the UAE; and major western economies, such as Canada and the US.

It was notable that the declaration was signed by both the US and China, for whom AI has become a lightning rod issue, economically and ideologically. There have been escalating tit-for-tat technology controls on both sides, with the US imposing fresh export controls on semiconductor chips earlier this year and China responding with graphite controls last month.

Speaking to the FT, UK secretary of state for science, innovation and technology, Michelle Donelan, described China’s presence as “monumental”. She also said:

“We have always said that no single country can face down the challenges and risks posed by AI alone.

“Bletchley Park marks the start of a long road ahead, and the summit will kickstart an enduring process to ensure every nation and every citizen can realise the boundless benefits of AI.”

In addition to China, other BRICS countries, such as Brazil and India signed the agreement, along with ASEAN nations: Singapore, Indonesia and the Philippines.

‘Significant risks’

The document outlines significant risks posed by AI, including towards human rights and “unforeseen risks stemming from the capability to manipulate content or generate deceptive content”.

Focusing on issues arising at the “frontier” of AI – the most advanced and untested iteration of the technology – it also identified “risks in domains such as cybersecurity and biotechnology” as being of particular concern.

To fully address “frontier” risks the declaration sets out two main strands of action. The first is “building a shared scientific and evidence-based understanding” of the risks and ensuring this is maintained as AI capabilities develop and collaborating on policies to address those risks.

The second strand recognises that “our approaches may differ based on national circumstances and applicable legal frameworks”. Within these differences, signatories have committed to common goals such as ensuring transparency within the private sector, with clear measures and tools for safety testing, alongside developing private sector capability.