ARIA Safeguarded AI Programme

ARIA Safeguarded AI Programme

Funds United Kingdom AI ventures with safeguarded technology programs for responsible models.

The ARIA Safeguarded AI Programme is an initiative of the Advanced Research and Invention Agency (ARIA), the UK's high-risk, high-reward research funding body established in 2023. The programme targets the development of mathematical and formal methods for guaranteeing safe and predictable behaviour in artificial intelligence systems, occupying a niche at the intersection of pure mathematics, theoretical computer science, and applied AI safety research. ARIA's mandate is to fund ambitious, speculative science that mainstream grant bodies are unlikely to back, making this programme structurally distinct from applied AI funding offered elsewhere in the UK innovation ecosystem.

The programme funds projects spanning TRL 2 through TRL 5, reflecting its focus on pre-commercial but technically substantive research. Award sizes range from approximately GBP 200,000 to GBP 5,000,000, with a representative median near GBP 1,200,000. The instrument is structured as a convertible rather than a straightforward grant, consistent with ARIA's design to retain optionality over promising outputs. Eligibility extends to both UK-registered organisations and international R&D performers, and for-profit companies are required applicants, though universities and research organisations may also participate. Sector restrictions limit scope to work on mathematical guarantees for AI safety.

Applications succeeding under this programme will typically demonstrate rigorous theoretical grounding, a credible pathway from formal proof methods toward practically deployable safety guarantees, and research leadership with track records in relevant mathematical or computational disciplines. Teams from UK universities, research institutes, and deep-tech startups operating at the research frontier are the primary intended beneficiaries.