UK Research and Innovation (EPSRC)

Next generation artificial intelligence (AI): explainable AI

UKRI funding for speculative, high-risk fundamental research that could bring a step change in how well future AI systems can be explained. It is the first opportunity under UKRI's IS8 AI Programme and particularly welcomes early and mid-career researchers at UK research organisations eligible for UKRI funding. Projects can cost up to £602,500 full economic cost, of which UKRI pays 80% (£482,000), must start on 1 February 2027 and last up to 24 months. Decisions are made by distributed peer review, so every applicant team must review around eight other proposals.

Apply on the funder’s site

You could get: £1 – £602,500
Closes in: 2 days left; 20 October
Effort: Heavy; budgets, references, often several stages

Who can apply

Standard UKRI eligibility: the project lead must be based at a UK research organisation eligible for UKRI funding, and that organisation submits the application
Early and mid-career researchers are particularly welcome
Joint leadership is allowed, with one agreed administrative lead
Each person can be involved in no more than two applications, and lead only one
International researchers can join only as project co-lead (international) under the UKRI-Research Council of Norway Money Follows Cooperation agreement or the UKRI-IIASA agreement; other international collaborators and UK partners not at eligible organisations take part as project partners
By applying, teams agree to review other proposals; failing to submit reviews on time means automatic rejection

What it funds

Potentially transformative, highly novel fundamental research with the potential to deliver a step change in the explainability of future AI systems, for example:

Why AI systems produce particular outputs, behaviours or decisions
Uncertainty quantification: assessing and communicating the reliability, confidence and limitations of AI outputs
Costs: contribution to the salary of the project lead and co-leads
Research, technical and project management staff
Research consumables including compute costs
Travel and subsistence, including the annual workshop and networking
Data preservation, data sharing and dissemination
Estates and indirect costs

projects are not expected to produce tangible outcomes such as new AI models or spinouts within two years, or to succeed in all their goals. Applicants must explain why the research cannot be funded through other routes such as responsive mode.

What it will not fund

Applications that consist of majority non-technical research such as purely social science, legal, policy or ethical studies
Incremental work lacking ambition to push the boundaries of AI explainability
Continuation of existing projects
PhD studentship costs
Equipment with a value over £25,000 per item (including VAT)
Uninvited resubmissions of projects submitted to UKRI or any other funder
No-cost extensions for reasons not related to people

How to apply

Make sure your research organisation is registered on the UKRI Funding Service (Je-S is not used)
Start an application on the UKRI Funding Service
Write a Vision and Approach document of no more than 3 sides of A4 and upload it as a single PDF
Answer the remaining questions, including applicant and team capability, and name the team members who will take part in distributed peer review
Your research organisation submits the application by 4pm on 20 October 2026
Review the around eight proposals allocated to your team by the review deadline

funding for this opportunity is subject to full business case approval. There is no project lead response or panel step.

Have ready

Vision and Approach document: up to 3 sides of A4, 11-point Arial or equivalent with 2cm margins, as a single PDF no bigger than 8MB, named with the Funding Service number
Letters or emails of support from each named project partner, as a single PDF, in English or Welsh

do not send letters of support from the host or project co-leads' research organisations.

More like this

Other grants open to organisations like yours.