Across the sector, AI adoption is growing but largely unstructured. The State of AI in Nonprofits 2025 Report reveals that 42% of nonprofits have only one or two staff members exploring AI, just 6% consider themselves AI experts and 76% have no formal AI strategy. This means that most organizations are figuring it out informally, and that is precisely what makes the conversation about transparency and accountability not just relevant, but urgent.
This is the other half of the story: how nonprofits can be transparent, accountable, and intentional in their use of AI, especially when it influences decisions that affect the communities we serve.
AI as a Productivity Tool
For most nonprofits, AI begins as a productivity tool, helping teams draft reports, produce policy briefs, and communicate more efficiently. The appeal is obvious: it allows small teams to punch above their weight. When AI assists with tasks that do not directly influence decisions about individuals, disclosure may not always be necessary. What matters most is that the output is accurate, readable and true to your organisation's values.
What we often overlook, however, is how far AI's role has extended beyond these routine tasks
AI as a Decision-Making Aid
Many organisations are increasingly turning to AI to identify potential grantees, profile stakeholders, allocate resources and screen beneficiaries. In such instances, AI may no longer be quietly assisting in the background; it could be actively shaping decisions that determine who benefit, who is seen and who is left behind.
For instance, as part of a grant mapping exercise, I used AI to identify organisations in West Africa which were undertaking capacity-development training internally. The goal was to find strong candidates for a particular funding opportunity. The AI delivered. It compiled a list of organisations with active websites, published reports, and a visible social media presence.
However, something was missing.
Some of the most impactful organisations working quietly in communities across the region did not appear on that list. Not because they were ineligible or didn’t have sufficient experience. But because they lacked digital visibility. No active website. No social media footprint. No published reports. To the AI, they were invisible and therefore non-existent.
This is the hidden bias of AI-assisted decision-making. It finds what is visible and surfaces what is searchable. However, in the development sector, some of the most meaningful work happens far from the digital spotlight in organisations that serve communities faithfully without ever trending online. When AI informs who gets considered for funding, those invisible organisations risk being systematically excluded not by human prejudice, but by algorithmic blindness. For most organisations, this is not an isolated incident; similar scenarios have played out in different decision-making spaces.
In my case, the initial outcome of my AI search made me pause and begin to review my results.
That pause was ethical. I began asking myself questions: who am I missing? How else can I find them? Who can double-check my list? Is my approach exhaustive enough? Those questions did not just improve my search; they made me accountable to the donor who trusted our methodology, to the organizations I could have missed, and to the communities they serve. Imagining those same organizations sitting across from me, asking how I found my beneficiaries, made one thing clear: accountability and transparency in this case matter, especially how we demonstrate them can strengthen or erode the trust our organisations have worked hard to build.
How can nonprofits therefore use AI in a way that maintains confidence in our methodology, honours our organizations' expertise and strengthens the trust our stakeholders have placed in us?
Building Transparency and Accountability in AI Use
The answer lies in intentional practice, built through the everyday choices we make about what we share, verify, document and disclose when AI is Involved. To build transparency and accountability, three areas demand particular attention: how AI intersects with organisational integrity and knowledge sharing, how it affects the trust your stakeholders place in you, and how oversight and ethical decision-making are handled when AI is part of the process.
1. AI should serve your mission, not compromise your integrity
When AI is being used within your organisation, one principle remains constant: verify before you publish or share. AI can produce confident, well-structured outputs that are factually incomplete, culturally biased or simply wrong. For nonprofits, credibility rests on the accuracy of their communications and the integrity of their programmes. Therefore, an unverified AI output is not just an error; it punches your reputation in the face. AI knowledge sharing within your organization starts with ensuring that staff understand not just how to use these tools, but how to evaluate their output in line with organisational standards and values.
2. Trust is your operational currency
For nonprofits, trust is not optional; it is your operational currency. It runs in every interaction, holding all stakeholder relationships together. It takes time to build, can be lost in an instant, and is remarkably difficult to restore once gone.
One area where trust is tested is how organizations handle data. People have the right to understand how their information is used, especially when AI is involved in processing it. When beneficiaries share their information with your organization, they are trusting you, not an algorithm. Therefore, being transparent about what data AI processes, how it is used,and who has access to it is not just best practice. It is how you tell your stakeholders that we see you, we respect you, and we will not let any tool make decisions about you without your consent.
For example, across West Africa, data protection is a legal obligation as well as an ethical responsibility. The ECOWAS Supplementary Act on Personal Data Protection establishes the regional framework, explicitly stating in Article 35 that 'no decision that has legal effect on an individual shall be based solely on processing by automatic means of personal data.' This principle speaks directly to how nonprofits use AI in decisions that affect beneficiaries, grantees and communities. National laws in Ghana and Nigeria reinforce these regional obligations further. Together, these frameworks reinforce the need for AI use to be transparent, accountable and subject to appropriate oversight, principles that organizations operating in the region should reflect in their AI practices.
3. The human must always be in the room
Oversight is the bridge between what AI recommends and what your organisation decides. This is the place where accountability lives. It may take many forms, including a simple question, an extra eye or a team review. AI can process data, identify patterns and generate recommendations faster than any human team. However, it cannot exercise moral judgement, understand the human consequences of its recommendations or take responsibility for outcomes.
AI can inform, but only humans can be accountable. Without oversight, AI takes charge, and no algorithm can answer to your board, donors or your communities.
From Principle to Practice
What, therefore, does responsible AI use look like in practice? Here are five ways that nonprofits, regardless of size and sector, can embed transparency and accountability in their use of AI.
1. Declare your data usage. Tell your stakeholders what data AI is processing, how it is stored, and who has access to it. Including a simple data usage statement in consent forms, programme guidelines, or on your website helps people feel safe about how their personal or organisational data is used. Always ensure compliance with the data protection laws applicable in your operating context. Take it a step further by drafting an AI policy for your organisation, a guide your team can rely on. Organisations developing their first AI policy can draw on free resources and policy guides available online, adapting them to their own context and needs. Hive Mind offers practical resources for civil society organisations, such as its guide Creating Your CSO’s Responsible AI Principles: A Practical Guide, which provides step‑by‑step approaches that can be tailored to local realities.
2. Disclose intentionally. Disclose intentionally. Not every task requires disclosure, but when AI has informed a decision that affects people, being transparent about it is not optional. In a grant report, for instance, a simple methodology note such as: 'AI-assisted tools were used in our initial stakeholder mapping, supplemented by direct community outreach and partner referrals' is clear, honest and professionally credible. Where AI plays a role in how resources are allocated, who is identified, or how decisions are made, make disclosure a consistent and standard part of your documentation. In a sector built on trust, that consistency is itself a signal of integrity. Transparency does not undermine your work; it strengthens it.
3. Go beyond the algorithm. When using AI to identify grantees, beneficiaries, or partners, supplement your search with grassroots outreach, partner referrals, and community networks. AI finds what is visible. Your job is to find what isn't. In one project, I used AI to identify potential interviewees. Then I cross-referenced the list with trusted contacts in our network to ensure no relevant stakeholders had been overlooked.
4. Verify before you act. AI outputs are a starting point, not a final answer. Before sharing, publishing or acting on anything AI generates, apply human judgement. Ask: Is this accurate, complete, and fair to everyone it affects? This is your basic and most powerful oversight mechanism and the foundation of ethical decision-making. Always have subject matter experts review AI outputs. Also verify outputs against trusted data sets, official reports, or relevant statistics to ensure credibility.
5. Invest in AI literacy. Transparency and accountability start with understanding not just how to use AI tools, but how to question them. Deep Learning Indaba and the African Academy of AI offer community learning and organizsation-level training rooted in responsible AI. To understand the legal and literacy frameworks shaping AI use in civil society, start with Hive Mind's Guidelines on the Challenges and Legal Framework of AI Literacy in CSOs. Beyond formal courses, also build simple habits such as setting aside time in team meetings to discuss which AI tools staff are using, what is working and what raises concern
Nonprofits are spaces of trust. The communities they serve entrust them with their data, stories, and vulnerabilities. As AI becomes increasingly embedded in nonprofit operations and decision-making, the values that guide this work must remain at the centre. Transparency, inclusion, and human oversight are essential in AI use to ensure that trust and accountability are safeguarded. While AI offers new possibilities, its use also raises difficult questions about whose voices are represented, whose interests are protected, and who bears the consequences when systems get things wrong. For nonprofits, responsible use of AI often begins with simple but important practices including asking the right questions, checking outputs critically, involving the people affected, and knowing when human judgement must take precedence.
Whether to use AI is no longer the question; it is already shaping how we work. Our responsibility as nonprofits is to use it in ways that reflect our values, protect our stakeholders, and honour the trust that makes our work possible. The communities we serve deserve nothing less.
Your Feedback Matters
What did you think of this text? Take 30 seconds to share your feedback and help us create meaningful content for civil society!
References
African Academy of AI. (n.d.). African Academy of AI. https://www.aaai.africa
Deep Learning Indaba. (n.d.). Deep Learning Indaba. https://deeplearningindaba.com
ECOWAS. (2010). Supplementary Act A/SA.1/01/10 on personal data protection within ECOWAS. Economic Community of West African States. https://www.statewatch.org/media/documents/news/2013/mar/ecowas-dp-act.pdf
Federal Republic of Nigeria. (2023). Nigeria Data Protection Act, 2023. Nigeria Data Protection Commission. https://ndpc.gov.ng/resources/
Google. (n.d.). Google AI essentials. Google Career Certificates. Google AI Essentials
HiveMind Community. (2024, May 16). Creating your CSO’s responsible AI principles: A practical guide. HiveMind. https://en.hive-mind.community/blog/1168
Lucas, J-F., & Lepinette, M. (2026). AI literacy and CSOs: Legal guidelines and the main challenges. Hive Mind. https://en.hive-mind.community/blog/1170,ai-literacy-and-csos-legal-guidelines-and-the-main-challenges
Republic of Ghana. (2012). Data Protection Act, 2012 (Act 843). Data Protection Commission. https://www.dataprotection.org.gh
Republic of Ghana. (2020). Cybersecurity Act, 2020 (Act 1038). Cyber Security Authority. https://www.csirt.gov.gh
TechSoup. (2023). AI usage policy guide for nonprofits. TechSoup. https://page.techsoup.org/hubfs/Downloads/AI-Usage-Policy.pdf
TechSoup. (n.d.). Generative AI usage policy template. TechSoup. https://page.techsoup.org/hubfs/Downloads/Template%20-%20Generative%20AI%20Usage%20Policy.docx
TechSoup & Tapp Network. (2025). The state of AI in nonprofits: Benchmark report on adoption, impact, and trends. TechSoup & Tapp Network. https://page.techsoup.org/ai-benchmark-report-2025
University of Helsinki & Reaktor. (n.d.). Elements of AI. https://www.elementsofai.com
Disclaimers
This piece of resource has been created as part of the AI for Social Change project within TechSoup's Digital Activism Program, with support from Google.org.
AI tools are evolving rapidly, and while we do our best to ensure the validity of the content we provide, sometimes some elements may no longer be up to date. If you notice that a piece of information is outdated, please let us know at content@techsoup.org.'
This content was researched and written by Angela Apedoh and edited with AI assistance.
About the Author
Angela Apedoh is an international development professional with experience supporting programme delivery and stakeholder engagement across Africa, Europe and Latin America. Working across diverse development contexts, Angela believes that emerging technologies, including AI, should be adopted responsibly to create value for the organisations and communities they are intended to serve. She is committed to contributing to the growing body of knowledge on the responsible use of AI to advance social change and development.