At in the present day’s AI Seoul Summit, Expertise Secretary Michelle Donelan introduced that the UK authorities will provide grants to researchers to check easy methods to shield society from AI dangers resembling deepfakes and cyberattacks.
Shahar Avin, an AI security researcher, and Christopher Summerfield, the UK AI Security Institute Analysis Director, will lead the programme throughout the UK authorities’s pioneering AI Security Institute.
The analysis programme will likely be delivered in partnership with UK Analysis and Innovation and The Alan Turing Institute.
Candidates have to be based mostly within the UK however are inspired to collaborate with different researchers from around the globe.
The worldwide impression of the AI Security Institute
The UK authorities’s pioneering AI Security Institute is main the world in testing and evaluating AI fashions, advancing the reason for secure and reliable AI.
Earlier this week, it launched its first set of public outcomes from exams of AI fashions.
It additionally introduced a brand new workplace within the US and a partnership with the Canadian AI Safety Institute – constructing on a landmark settlement with the US earlier this yr.
Expertise Secretary Michelle Donelan mentioned: “When the UK launched the world’s first AI Security Institute final yr, we dedicated to reaching an formidable but pressing mission to reap the optimistic advantages of AI by advancing the reason for AI security.
“With analysis techniques for AI fashions now in place, Part 2 of my plan to soundly harness the alternatives of AI must be about making AI secure throughout the entire of society.”
Mitigating the potential impacts of AI
The brand new grants programme is designed to broaden the Institute’s remit to incorporate the rising discipline of ‘systemic AI security’.
This discipline goals to know easy methods to mitigate AI’s impacts at a societal degree and examine how our establishments, techniques, and infrastructure can adapt to the transformations this expertise has led to.
Examples of proposals inside scope would come with concepts on easy methods to curb the unfold of faux photographs and misinformation by intervening on the platforms that unfold them somewhat than on the AI fashions that generate them.
Christopher Summerfield, UK AI Security Institute Analysis Director, commented: “This new programme of grants is a serious step in direction of guaranteeing that AI is deployed safely into society.
“It’s designed to generate an enormous physique of concepts for easy methods to deal with this drawback and to assist make certain nice concepts may be put into apply.”
