In 2026, AI funding will only be granted to projects with demonstrable development risk. The EU AI Act, which is being phased in starting 2025, is establishing itself as the baseline for eligible solutions.
Funding providers assess AI projects against three key criteria to determine eligibility: Is there genuine technical development risk? Does the planned solution go beyond standard software available on the market? Is independent research and development (R&D) being performed? The dividing line between implementation and funding-worthy R&D is often narrow. Integrating a standard API into an existing database is classified as pure implementation, while training on proprietary, unstructured data to meet specific accuracy specifications that standard solutions cannot achieve represents a development risk and is therefore eligible for funding.