$11M in AI grants aim to help states modernize public benefits

States will receive millions in new funding to experiment with artificial intelligence tools aimed at modernizing public benefits administration as they prepare for sweeping federal changes to Medicaid and the Supplemental Nutrition Assistance Program under H.R.1. 

The Center for Civic Futures, a program by Renaissance Philanthropy, announced Thursday that it’s awarding $11 million to 11 projects through its Public Benefit Innovation Fund. The projects will explore ways to reduce administrative burdens, improve eligibility processing and update technology infrastructure behind benefits programs, including Medicaid, SNAP and unemployment insurance.

The announcement comes as states face new federal requirements for benefits eligibility, work reporting and payment accuracy. H.R.1, the budget reconciliation law enacted in 2025 and widely known as the One Big Beautiful Bill Act, requires states to prepare for more frequent Medicaid eligibility checks and new work requirements for certain beneficiaries, while shifting more responsibility for SNAP administrative costs and payment errors to states. 

Cassandra Madison, executive director of the Center for Civic Futures, said responses to the organization’s first funding round last year, which focused broadly on ideas for addressing H.R. 1 changes, revealed a need to direct more investment toward the operational challenges government agencies face. 

“Close to half of the applications that came in were [to develop] chatbots,” Madison told StateScoop, which, she said, reflected a mismatch between what developers want to build and what government agencies need.

“If you are a builder and want to build something for the public good, but you don’t really understand what the day-to-day struggles of government are, that’s the easiest thing to go after,” she added.

This new round of funding, Madison explained, emphasizes the behind-the-scenes processes, including work caseworkers perform to verify eligibility and coordinate information across government systems. 

“In all of our conversations with states, we very purposely landed an open call that was much more backend-focused,” Madison said. “If you think about what the average caseworker does and how much time they spend on paperwork instead of actually supporting a human one-on-one, we have to go after that administrative burden.”

Eleanor Davis, who serves as director of the Public Benefit Innovation Fund and worked on developing the second round of projects for CCF, said the civic technology field has historically concentrated on improving the experience of people applying for benefits, while giving less attention to the systems government employees use to process applications. 

“There’s infinite reasons why that backend work is really hard to do,” Davis said in an interview. “And now is a critical moment to be driving innovation towards where we can leverage AI to make a huge difference in these spaces where we have not had as much opportunity to improve service delivery.” 

Davis said the funded projects will address streamlining eligibility processing, making better use of government data and tackling underlying barriers such as outdated technology, procurement and workforce shortages.

Among the projects, North Carolina will test an AI-assisted process for recertifying SNAP benefits, initially using synthetic data to evaluate the approach before rolling out an official launch. Stanford RegLab will work with Maryland and Colorado to identify missing or unclear information in unemployment insurance applications. Other projects will explore identifying SNAP cases likely to contain errors, connecting corrections data with Medicaid and SNAP systems, and using AI to map legacy technology systems more quickly. 

The grants will also support efforts to help Medicaid agencies develop more flexible technology contracts and translate SNAP eligibility rules into transparent, computer-readable formats — what Davis calls “upstream barriers.”

“These barriers that have just been persistent blockers to good service delivery for decades,” Davis said. “You can’t test and launch an amazing new AI tool that’s going to solve all your problems if you can’t even buy it.”

The grants vary in length from six months to two years. Some will remain in controlled testing environments, while others are designed for pilots involving real-world systems and users. 

Having a closed sandbox, Madison explained, is especially important when experimenting with tools that rely on sensitive data — including names, Social Security numbers, dates of birth, home addresses and phone numbers— which Medicaid and SNAP both use to verify program eligibility.

“These technologies are aimed at supporting people,” Madison said. “Not only do we have personally identifiable information, but this is some of the most vulnerable times in people’s lives.”

Davis said success will depend on the specific project, with teams evaluating outcomes such as reduced payment error rates, improved data processing and more efficient casework workflows. Even if an AI tool isn’t successful, she said they consider it just as valuable to discover if an approach doesn’t work as expected.

“I think the idea of success is really learning for us,” she said. “Everything will be open source to be able to share those lessons, so that other states can take those tools and scale them for national impact.”

The new funding opportunity also builds on efforts from states across the country to use AI to respond to H.R.1. 

Earlier this year, Code for America identified several AI use cases for Medicaid service delivery, including conversational interfaces, data quality analysis, process automation, and risk and fraud detection. Danny Mintz, director of safety net policy at CFA, said agencies should first conduct pilot studies and develop evaluation criteria and success metrics to ensure these tools are effective and do not exacerbate health disparities between populations.

“The real criteria are how this tool performs for people with different demographic characteristics and is there something about a particular tool that seems to be systematically disadvantageous to a particular segment of the population who’s receiving that intervention,” Mintz said in a January interview.

In May, CFA announced plans to integrate Anthropic’s Claude into SNAP caseworker tasks like reviewing eligibility documents and answering policy questions.

In September, Maryland partnered with civic technology organizations to simplify Medicaid applications and notices as well as identify respective caseworker tasks that could be automated. 

The Center for Civic Futures said its first funding round has already produced early results. A SNAP verification tool in Mississippi helped reduce the state’s payment error liability by $40 million, while AI tools in New Jersey have helped deliver an estimated $19 million in benefits to more than 97,000 residents. 

Despite the pressures state agencies face, Madison said she has seen growing interest among government leaders to rethink service delivery.  

“Even in the midst of really challenging circumstances, increased administrative burden, staffing shortages, budget shortages, we see so much potential and energy among leaders when you put them in a room,” Madison said. “Never would I have imagined this in government a decade ago.”

Davis described state human services leaders as “determined” to use the tools available to them to help eligible residents access food assistance and health care. 

“Between the biggest policy changes in a generation, the pressures of technology and workforce and budget …there’s so many hard things that they are navigating,” Davis said. “This technology actually has the potential to truly unlock things and tackle problems that have not been able to move in decades.”