Munim Technologies

Research/Society

How much federal legal-aid tech money now goes to AI

AI projects went from 1 of 35 Legal Services Corporation technology grants in 2021 to 12 of 32 in 2025, and from 2.2% of the dollars to about 40%. Programs in seven states received no grant in any of the five rounds.

Oct 14, 2026 · 6 min read

The Legal Services Corporation's Technology Initiative Grants (TIG) are the main federal source of technology money for civil legal aid in the United States. We built the first machine-readable list of every award from 2021 to 2025 and flagged each one whose description mentions AI. AI went from 1 of 35 awards in 2021 ($93,290, 2.2% of dollars) to 12 of 32 in 2025 ($1,643,948, 40.4%). In 2025, eight awards put AI into intake, the first conversation a person in trouble has with a legal aid office. Meanwhile, LSC-funded programs in seven states received no TIG at all in those five rounds.

2.2% → 40.4%

AI share of listed TIG dollars, 2021 to 2025

12 of 32

2025 awards whose description mentions AI

$1,046,054

held by 8 AI intake awards in 2025, a quarter of that round's dollars

7 states

with no TIG in any round from 2021 to 2025

The question

Civil legal aid helps people with evictions, debt, benefits and family cases. Demand far exceeds the number of lawyers, and AI is often proposed as a way to close the gap. How much of the federal money for legal aid technology is actually going to AI, and where does it go? LSC publishes each year's awards as a press release or web page, but we found no machine-readable list of the 2021-2025 rounds and no published count of AI awards or AI dollars.

What we did

We copied every award LSC printed for the 2021-2025 rounds, one record per award, in two independent passes that gave the same names and amounts. An award counts as AI when LSC's own project description names AI: the word "AI", "artificial intelligence", "generative", "chatbot" or an AI product such as Microsoft Copilot. "Automated" alone does not count. Each AI record keeps the fragment of LSC's text that triggered the flag, so every call can be checked. A separate flag marks descriptions that mention intake, screening, triage or a hotline.

What we found

The rise is steady in the count of awards and steep in dollars. AI projects took 21.1% of listed dollars in 2023 and 43.1% in 2024. In 2025, 40.4% of a smaller round went to AI. In 2024, nine AI awards, 28.1% of the count, took $2,720,420, 43.1% of the dollars.

AI's share of Technology Initiative Grants

Share of dollars listedShare of awards
01020304020212022202320242025TIG roundShare (%)
An award counts as AI when LSC's description names AI. One 2025 AI award was printed with no amount and is left out of the dollar shares. Source: lsc-tig-awards/data/awards.json (README table, generated by tig.py)

TIG awards and AI awards by round

RoundAwardsMention AIDollars listedAI dollarsAI and intake
2021351$4,268,931$93,2900
2022332$4,679,135$261,4181
2023334$5,157,033$1,089,6341
2024329$6,306,081$2,720,4202
20253212$4,064,308$1,643,9488
Source: lsc-tig-awards/data/awards.json; recomputed and matching the README table.

AI intake arrived in 2025

Only four AI intake awards came in the four rounds from 2021 to 2024. The December 2025 round had eight. They range from an AI voice hotline in rural Florida to agentic AI completing phone intakes in Middle Tennessee and generative AI in Minnesota's statewide intake. Every TIG requires an evaluation plan, so these projects will be measured in 2026 and 2027.

The eight 2025 awards that put AI into intake

GranteeStateAmountLSC's wording
Florida Rural Legal ServicesFL$164,438an AI-powered hotline
Legal Aid Society of Middle Tennessee and the CumberlandsTN$209,056integrate agentic AI
Southern Minnesota Regional Legal ServicesMN$239,360integrate generative AI
Legal Aid Services of OregonOR$250,000automated scheduling and potential AI features
Legal Services of Eastern MissouriMO$78,200AI-driven intake system
Legal Aid of the BluegrassKY$35,000a written implementation plan detailing AI integration
Legal Services of the Hudson ValleyNY$35,000multilingual, AI-supported technologies
Legal Aid of Southeastern PennsylvaniaPA$35,000a unified client portal and AI features
Source: lsc-tig-awards/data/awards.json (mentions_ai and mentions_intake both true, year 2025; ai_terms quoted). Total $1,046,054.

The wording shows how wide the AI flag is. "Potential AI features" counts the same as a project built around a model. That is a limit of using LSC's one-paragraph summaries, and it is why every record carries its quote.

Where the money does not go

TIG money is won by competition, and it is unevenly spread. LSC-funded programs in Indiana, Iowa, Maryland, New Jersey, Rhode Island, Utah and Wyoming received no TIG in any of the five rounds, and neither did any program in a US territory. For Iowa, Maryland and Rhode Island, LSC's own pages show no award in 2018, 2019 or 2020 either: eight rounds in a row. At the other end, programs in eight states won in every round.

States by number of 2021-2025 rounds with at least one TIG

50 states and DC (DC counted as a state). Territories, which also had no award, are not included. Source: lsc-tig-awards/README.md (table generated by tig.py --states from data/awards.json)

Set against the poverty-based money LSC gives each state, the gap is wide. Over the five rounds, TIG money came to 58 cents per person in poverty across the 56 jurisdictions with LSC basic field funding. Vermont received $12.39 per person in poverty, about 300 times Virginia's 4 cents. The three largest are far below their share: California holds 10.8% of the people in poverty and received 4.7% of TIG dollars, Texas 10.1% and 5.6%, New York 6.4% and 1.2%.

What it means

Federal legal-aid technology money has shifted quickly toward AI, and in 2025 it shifted toward the front door: the intake call. That makes the coming evaluations important. If AI intake works, the programs that got none of this money are the ones least able to build it themselves. The data cannot say why a state had no award; LSC publishes awards, not applications.

More research