The Double Digital Divide

We know that the digital divide harms marginalized communities, but what happens when it also strikes the institutions that serve them?

AI is spreading quickly through the American economy, but its benefits and risks are not landing evenly. At Nonprofit AI Studio, we call this the double digital divide: the communities with the least access to AI are served by the nonprofits with the least capacity to adopt it safely. The two gaps compound each other, which makes this an equity issue, not only a technology one.

1. Marginalized communities have less access to AI. AI is becoming a path to opportunity, and jobs requiring AI skills pay an average premium of 62% [1]. Yet only 41% of lower-income adults use AI chatbots, compared with 66% of upper-income adults [2], and only 54% of adults in households earning under $30,000 have home broadband, compared with 94% of those earning over $100,000 [3]. Language is another barrier. Nearly 28 million U.S. residents have limited English proficiency [4], and tests of AI tools have found weaker translations in languages such as Arabic, Farsi, and Pashto, including errors in health care, domestic violence, and legal content [5].

2. They are served by small, community-led nonprofits. Most nonprofits are small, and 92% operate on less than $1 million a year [6]. The organizations closest to low-income families are disproportionately led by people of color: in one national survey, 83% of BIPOC-led nonprofits primarily served people with lower incomes, compared with 62% of white-led nonprofits [7]. CEOs of color lead 46% of very small nonprofits but 24% of others [8], and women lead 58% to 62% of nonprofits under $500,000 but just 31% of those over $50 million [9].

3. Those nonprofits face the same barriers to AI. AI is already changing what nonprofits can deliver. Among nonprofits that have built AI into their programs, 92% say it made service delivery more efficient and 41% use it to reach people in new languages and formats [10]. But nonprofits with budgets over $1 million adopt AI at nearly twice the rate of smaller ones, 66% compared with 34% [11], and funders are not filling the gap. Ninety percent of foundations offer no funding or other support for grantees' AI use [12], and just 22% of very small nonprofits received a foundation grant between 2019 and 2023, compared with 72% of other nonprofits [8]. That leaves AI to be paid for with unrestricted money, which most often comes from individual donors, and 63% of leaders of color lack access to individual donors, compared with 49% of white leaders [13].

4. When they do adopt AI, it's often without protection. Only 27% of nonprofits under $10 million have written guidance on safe AI use, compared with 56% of larger ones, and more than half of nonprofit professionals in the same survey had used AI for work informally or outside any guidance [14]. That matters because nonprofits hold sensitive information, from donor financial records to client health, legal, and immigration details [15], and because the public trusts nonprofits more than any other U.S. institution [16].

Falling behind together. These gaps do not simply sit side by side; they compound. Without AI capacity, the nonprofits serving low-income, BIPOC, immigrant, and LGBTQ+ communities either don't adopt AI, which leaves them less efficient and less competitive for funding, or adopt it informally, which puts their clients' data at risk. As those communities lose ground, their needs grow and land on organizations that are already stretched. Without deliberate investment, communities and the nonprofits that serve them will sink further behind together, and AI will widen existing inequities in who gets served, how well, and how safely.

This summary draws on our full white paper, The Double Digital Divide: How the AI Gap Affects Marginalized Communities and the Nonprofits That Serve Them, which includes more data, notes on methodology, and known data gaps. Much of the research on nonprofits and AI comes from voluntary surveys that underrepresent the smallest organizations, so where a statistic comes from a specific sample, the white paper says so.

Sources

1. PwC, 2026 Global AI Jobs Barometer

2. Pew Research Center, Americans and AI 2026: Chatbots, Smart Devices and Views on Impact

3. Pew Research Center, Internet use, smartphone ownership and digital divides in the U.S. (January 2026)

4. Migration Policy Institute, language access event summary

5. Respond Crisis Translation, Lost in AI mistranslation (2025)

6. National Council of Nonprofits, Nonprofit Impact Matters

7. Nonprofit Finance Fund, 2022 State of the Nonprofit Sector Survey: A Focus on Racial Equity

8. Candid, The 'invisible majority': What we know about very small nonprofits (2025)

9. Candid, 2026 Nonprofit Compensation Report press release

10. Fast Forward, 2026 AI for Humanity Report

11. NonProfit PRO, State of AI in Nonprofits 2025 (TechSoup and Tapp Network)

12. Center for Effective Philanthropy, AI With Purpose (2025)

13. Giving Compass, Supporting Nonprofit Leaders of Color (citing Building Movement Project)

14. NTEN and the Bridgespan Group, 2026 State of Nonprofit AI

15. Meet the Moment, The Cybersecurity Landscape course for NTEN

16. Independent Sector and Edelman Data & Intelligence, Trust in Nonprofits and Philanthropy (2026)