US fund awards $11m to test AI in public benefits and government services

A new $11 million funding round will support 11 projects across the United States testing how artificial intelligence can improve access to public benefits, reduce administrative errors and modernise the technology used by government agencies The funding, announced by the Center for Civic Futures, forms the second round of its Public Benefit Innovation Fund. Recipients include university researchers, state governments, nonprofits and technology companies working on services such as food assistance, healthcare coverage and unemployment insurance. The programme is backed by Ballmer Group, the Gates Foundation, Recoding America and Arnold Ventures. OpenAI and Anthropic will also provide technical assistance to participating teams. The projects will examine specific problems in public service delivery, including complicated eligibility rules, outdated computer systems and the administrative workload involved in processing applications and checking claims. Universities and state agencies to test AI in benefits processing Several projects will examine whether AI can help government agencies process applications more accurately while reducing the demands placed on caseworkers and applicants. At Stanford RegLab, researchers will work with Maryland and Colorado to develop and test an AI tool for unemployment insurance applications. The system will attempt to identify missing or unclear information during the application process, allowing problems to be addressed earlier. North Carolina is developing a separate approach to recertifying eligibility for the Supplemental Nutrition Assistance Program (SNAP), the US federal food assistance programme. Its project will target administrative steps that frequently cause delays or errors. The state plans to use synthetic data during its initial testing, allowing the technology to be evaluated before any live deployment. At Yale University, the SCALE Lab and Better Government Lab will work with five states to develop predictive models that identify food assistance cases most likely to contain errors. The researchers aim to improve the accuracy of case reviews without creating additional paperwork for families. Another project, led by Mirza, will investigate how existing medical claims and health records could help identify Medicaid recipients who qualify for exemptions from work requirements because of medical conditions or disabilities. The work comes as US states prepare for changes to federal benefits requirements, increasing pressure on agencies to verify eligibility while maintaining access for people entitled to support. Axiom Foundation to make benefits eligibility rules more transparent One of the selected projects will examine a problem that sits behind many automated government decisions: how complex legislation is translated into computer systems. The Axiom Foundation will work with a US state agency to test an open, machine-readable version of SNAP eligibility rules. These rules determine who qualifies for food assistance, but are frequently implemented separately within government software and other eligibility tools. Axiom's approach involves converting written legislation and policy requirements into executable code, with each rule linked to its original legal source. The aim is to make the calculations used by benefits systems easier to inspect, verify and update, while providing a consistent foundation for AI applications. Ariel Kennan, who shared details of the project on LinkedIn, said: "We will work with a state agency to test open encodings of SNAP eligibility rules." The foundation's wider work builds on PolicyEngine infrastructure, which has been used to model tax and benefit policies in the US and UK. A separate project from Digital Public Works will examine income verification. Its proposed platform would allow state agencies to connect to multiple verification providers and receive information through a single interface rather than managing different reporting systems. AI projects target ageing government technology and caseworker workloads Other recipients will focus on the systems and processes that make public benefits difficult to administer. Technology organisation Nava will test whether AI can shorten the process of analysing legacy government software, which can take months before modernisation work begins. The project aims to help agencies understand existing systems more quickly and develop plans for replacing or updating the technology supporting Medicaid and food assistance services. Foster Insights will investigate whether AI can identify unmet family needs by examining caseworker notes, including possible requirements for housing, childcare and financial support. Recidiviz will develop a system connecting information held by corrections agencies with Medicaid and SNAP administration. The project is intended to help people leaving prison retain access to benefits for which they remain eligible. Meanwhile, Holly and 17A will work with two state governments to examine how AI may change public sector jobs, the skills employees require and the way agencies organise their workforces. The Open Contracting Partnership will focus on technology procurement, developing shared resources to help state Medicaid agencies purchase systems that can be updated more easily and evaluated against service outcomes. OpenAI and Anthropic to support technical development OpenAI and Anthropic are contributing technical resources to the latest funding round as participating teams develop and evaluate their projects. The companies will provide AI credits, weekly sessions with applied AI engineers and monthly technical workshops covering system architecture, AI evaluation, agent design and implementation. The support will allow recipients to work directly with engineers from both companies while testing their proposed applications. Cassandra Madison, Executive Director of the Center for Civic Futures, said the funding was shaped by the operational problems identified by government agencies. "For too long, government has had to adapt to available technologies instead of shaping what gets built. It's time to rethink that equation." Fund reports early results from previous AI projects The Public Benefit Innovation Fund launched in 2025, initially committing $8.5 million to seven projects addressing changes to public benefits administration. The Center for Civic Futures says that work is already producing measurable results. In Mississippi, a SNAP work requirement verification and reporting tool developed by Promise helped the state reduce its payment error liability by $40 million, according to the organisation. In New Jersey, AI-supported tools have helped deliver an estimated $19 million in public benefits to more than 97,000 residents. With the latest awards, the Center for Civic Futures has committed nearly $20 million across 18 projects. The new funding round will extend testing into areas ranging from benefits eligibility and caseworker support to public sector workforce planning and the underlying software that government agencies depend on. ETIH Innovation Awards 2027 open for entries The ETIH Innovation Awards 2027 are now open for entries, with 26 categories recognising companies, institutions, projects and individuals making a measurable impact across education technology. The global awards span AI in education, digital learning, higher education, skills and workforce, inclusion, education operations, evidence and impact, with entries managed through ETIH’s new online awards platform. Entries are open worldwide, with the early-bird deadline on Friday 18 December 2026 and the final entry deadline on Friday 26 February 2027 at 23:59 UK time. Explore the 2027 categories and judging criteria, read the How to Enter guide or start an entry through the ETIH Innovation Awards 2027 platform.