Grant scheme to put AI safety at ‘the heart’ of government

The UK Government is to make £8.5m of research grants available to enhance society’s resilience to the risks posed by artificial intelligence (AI).

UK technology secretary Michelle Donelan announced the funding at the AI Seoul summit, an event being co-hosted by the UK and the Republic of Korea. 

Donelan said: “With evaluation systems for AI models now in place, phase two of my plan to safely harness the opportunities of AI needs to be about making AI safe across the whole of society. 

“This is exactly what we are making possible with this funding, which will allow our institute [the AI Safety Institute] to partner with academia and industry to ensure we continue to be proactive in developing new approaches that can help us ensure AI continues to be a transformative force for good.”

Successful applicants will work to tackle issues such as deepfakes and cyber-attacks as well as harness AI's opportunities.

Donelan added: “I am acutely aware that we can only achieve this momentous challenge by tapping into a broad and diverse pool of talent and disciplines, and forging ahead with new approaches that push the limit of existing knowledge and methodologies.”

Examples of proposals could include ideas to stop the spread of fake images by intervening on the platforms that spread them, rather than on the AI models that generate them.

The AI Safety Institute (AISI) will lead the programme, with help from the UK Research & Innovation Institute and The Alan Turing Institute. 

Helen Margetts, director of public policy at The Alan Turing Institute, said: “Rapidly advancing technology is bringing profound changes to the information environment, shaping our social, economic, and democratic interactions. 

“That is why funding AI safety is vital to ensure we are all protected from the potential risks of misuse while maximising the benefits of AI for a positive impact on society.”

AISI will also look to collaborate with fellow institutes across the world as part of its plans to boost cross-border working. Earlier this month it launched a new office in the US and a partnership with the Canadian AI Safety Institute.

Christopher Summerfield, AISI research director, who will co-lead the programme, said: "This new programme of grants is a major step towards ensuring that AI is deployed safely into society.

"We need to think carefully about how to adapt our infrastructure and systems for a new world in which AI is embedded in everything we do. 

"The programme will bring safety research right into the heart of government, underpinning the pro-innovation regulation that will shape the UK's digital future."