The UK Government’s AI Security Institute has launched its first round of grant funding for reserch projects under the £5 million Challenge Fund programme.
It said the fund gives researchers a chance to obtain up to £200,000 for projects to harden critical industries, prevent the misuse of AI, bolster public confidence in the technology and promote its adoption across the economy.
The institute – which recently changed its name to place the emphasis on security – has called on researchers to submit proposals and said it will make a decision within 12 weeks of submission.
It wants to build a strong evidence base to understand and mitigate the most serious threats posed by advanced AI systems and ensure the UK’s critical infrastructure is protected.
The institute’s guide to the fund specifies four priority areas for research: safeguards against the misuse of AI; the design and evaluation of control protocols; alignment of AI’s goals and actions with human intentions; and building resilience into systems and infrastructure.
Security, resilience, trust
Minister for AI and Digital Government Feryal Clark said: “AI is at the heart of our Plan for Change - driving economic growth, creating jobs, and transforming public services for people across the country. But to unlock its full potential, we must ensure AI systems are secure, resilient, and trusted – with safety baked in from the start.
“This fund supports world class research to tackle the toughest safety and security challenges in AI, protecting critical infrastructure and removing barriers to adoption. By addressing these challenges head on, we’re laying the foundations for AI to boost productivity, strengthen public services and power a decade of national renewal.”
AI Security Institute chair Ian Hogarth said: “Making sure AI systems are aligned and operate with human oversight are two of the key open questions in technical AI safety.
“By funding high impact research across these priority areas, we’re building the evidence base needed to develop a robust understanding of, and real world solutions for, the most urgent security risks AI presents.”
