ARIA is a UK government research funding agency that backs high-risk, high-reward R&D in underexplored areas, including a major £59 million programme on formal mathematical safety guarantees for AI systems.
ARIA is a UK government research funding agency that backs high-risk, high-reward R&D in underexplored areas, including a major £59 million programme on formal mathematical safety guarantees for AI systems.
People
Updated 05/18/26Chief Technology Officer
Chief Executive Officer
Programme Associate (via Pace)
Senior Executive Assistant to Deputy CEO & CoS
Programme Director
Consultant
Talent Acquisition Manager
Chair
Funding Details
Updated 05/18/26- Annual Budget
- $27,600,000
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26The Advanced Research + Invention Agency (ARIA) was formally established on 25 January 2023 under the Advanced Research and Invention Agency Act 2022, following years of advocacy for a UK equivalent of DARPA. It operates as a non-departmental public body sponsored by the Department for Science, Innovation and Technology, with accountability to Parliament but significant operational independence to pursue bold, long-term research goals. ARIA's mission is to fund breakthrough R&D in underexplored areas to catalyze new paths to prosperity for the UK and the world. The agency operates through Programme Directors — scientific and technical leaders who define opportunity spaces and shape research directions. As of early 2026, ARIA had 16 Programme Directors across two cohorts, 53 staff, over 200 funded research teams and collaborators, and over 100 active projects. It has committed over £400 million to the UK R&D ecosystem since its founding. ARIA's founding CEO was Ilan Gur. In February 2026, Kathleen Fisher became CEO; she previously led DARPA's Information Innovation Office, overseeing a $500 million portfolio and over 50 programmes. Matt Clifford serves as Chair and Ant Rowstron as Chief Technology Officer. The agency's most directly AI-safety-relevant programme is Safeguarded AI, a £59 million initiative within the Mathematics for Safe AI opportunity space. Led by Programme Director David 'davidad' Dalrymple, Safeguarded AI aims to develop quantitative safety guarantees for AI systems using formal mathematical proofs and scientific world models — an approach inspired by safety certification standards in nuclear power and aviation. The programme operates across three technical areas: TA1 builds formal platforms for world models and proof verification; TA2 uses frontier AI to help domain experts create verifiable mathematical models of real-world dynamics; and TA3 applies these tools in real-world safety-critical contexts. In 2025 the programme pivoted away from planned cyber-physical applications toward cybersecurity demonstrations, responding to rapid advances in frontier AI capabilities. A second AI-relevant programme, Scaling Trust (approximately £50 million, under the Trust Everything, Everywhere opportunity space), seeks to enable AI agents to securely coordinate, negotiate, and verify with one another on behalf of humans. Phase 1 funding applications closed in March 2026. ARIA's broader portfolio includes programmes in synthetic biology, neurotechnology, climate science, robotics, and ecological engineering. Its government funding commitments extend through at least 2030, with £1.22 billion allocated for 2027-2030 in addition to the original five-year £800 million envelope.
Theory of Change
Updated 05/18/26ARIA believes that transformative scientific and technological breakthroughs require funding for research that is too speculative, too hard, or too interdisciplinary for mainstream institutions to support. For AI safety specifically, ARIA's theory is that empirical testing alone is insufficient to guarantee the behavior of advanced AI systems operating beyond human experience, and that mathematical proof-based approaches — formal verification of AI against rigorously specified world models — can provide the quantitative safety guarantees needed for trustworthy deployment in critical domains. By funding early-stage foundational work in formal verification, world modeling, and AI security that the private sector and traditional academic funders underinvest in, ARIA aims to establish a technical infrastructure that allows advanced AI to be deployed responsibly at scale, reducing the risk of catastrophic failures in safety-critical systems.
Grants Received– no grants recorded
Updated 05/18/26Projects
Updated 05/18/26Discussion
No comments yet. Be the first to share your thoughts.