UK to set up world’s first AI safety institute

British Prime Minister Rishi Sunak has unveiled plans to create the world’s inaugural Artificial Intelligence (AI) safety institute. The institute will be tasked with thoroughly assessing and testing new AI models to understand their capabilities and potential risks, from social biases to more extreme concerns.

Sunak is convening a global summit on November 1-2 at Bletchley Park, bringing together AI companies, political leaders, and experts to address the perceived existential risks associated with AI. The aim is to foster international consensus on its safe development. Notable attendees at the summit include US Vice-president Kamala Harris and Google DeepMind CEO Demis Hassabis.

Sunak envisions the UK as a leading force in AI safety, positioning itself strategically post-Brexit between the US, China, and the EU economic blocs. The summit will gather around 100 participants to explore topics including the unpredictable progress of AI and the potential for human loss of control over it.

While acknowledging AI’s potential to drive economic growth and address once-insurmountable challenges, Sunak emphasised the need to confront the new risks and fears it brings. He assured the public of the government’s commitment to ensuring their safety while harnessing the opportunities for a better future that AI offers.

The UK government has released a report on “frontier” AI, focusing on cutting-edge general-purpose models, which will guide discussions on risks related to societal impact, misuse, and loss of control.