ARIA Safeguarded AI: Cybersecurity, 31 October 2026 batch
ARIA funds teams that use AI and formal proofs to build secure software, with £2.5 to 3.5 million per Blue Team. The next batch closes on 31 October 2026.
The UK research agency ARIA has reopened its Safeguarded AI: Cybersecurity funding call on a rolling basis. The call is backed by £20 million. It funds teams that build security critical software components whose key security properties are backed by machine checked proofs, using AI to reach a level of formal proof that would otherwise not be possible.
There are two tracks. Track 1 funds 3 to 6 Blue Teams with about £2.5 to 3.5 million each, under ARIA's R&D funding terms. Track 2 funds one central Red Team with about £2 to 3 million under commercial service terms. The Red Team tries to break every Blue Team's system in each 8 week sprint. Projects run 10 to 15 months, until the end of November 2027.
ARIA welcomes applications from startups, industry and academia. It typically requires most of the project work to happen in the UK, meaning more than 50% of project costs and personnel time. Exceptions are possible where a team makes a strong case for the benefit to the UK. Teams may apply to both tracks, but one team will not be appointed to both.
Applications are open now and are reviewed in quarterly batches. The first cycle opened on 4 September 2026. Proposals submitted by 31 October 2026 at 14:00 GMT enter the November review, and applicants hear back on 30 November 2026. Later proposals move to the next cycle. ARIA says it may change the dates, pause the call or close it early.
Apply through the ARIA grant portal linked from the call page. Track 1 proposals are up to 5 pages and Track 2 proposals up to 3 pages. ARIA looks for teams with deep expertise in security and systems engineering, formal methods and proof engineering, and AI assisted workflows, who can work in short cycles and have a credible path to real world deployment.