AI is already reshaping how social impact organizations work. Adoption continues to rise across the sector, while governance often lags behind. Recent research shows that 82% of nonprofits are using some form of AI, yet 76% still do not have an AI policy in place. That gap creates both risk and missed opportunity for organizations working to advance their missions.

IBM and NationSwell developed Responsible Use of AI for Social Impact to help address that challenge. The playbook offers practical guidance for nonprofits, funders, universities and technology partners seeking to adopt AI in ways that are ethical, safe and useful in real-world settings.

The playbook focuses on the organizational capabilities needed to use AI responsibly, rather than treating AI as a standalone technology decision. Three areas sit at the center of that approach:

• Foundational AI literacy: Staff and leaders need core knowledge to understand AI tools, limitations and practical use cases.
• Equitable, human-centered design: Organizations need guidance on building around community needs and outcomes, rather than around technology for its own sake.
• Responsible development, deployment, and governance: Successful adoption requires clear approaches to transparency, fairness, privacy and internal oversight.

The playbook frames responsible AI adoption as a cross-sector effort, not something nonprofits should have to navigate alone. The whitepaper outlines the roles different stakeholders can play across the sector: nonprofits as mission drivers, universities as talent builders, technology partners as implementation experts and funders as capacity enablers.

“Even smaller nonprofits can use generative AI to automate repetitive administrative tasks, so staff can focus on high-value tasks like relationship building,” said Lydia Logan, Vice President, Global Education and Workforce Development, IBM.

Funders, in particular, have an important role to play. The whitepaper highlights two significant opportunities: investing in shared infrastructure for public-purpose AI and supporting AI literacy and capacity building across grantee organizations. That kind of support can have measurable impact. As noted in the whitepaper, nonprofits that build AI fluency may see efficiency gains of 10-15%, equivalent to a $100,000 unrestricted grant for a $1 million nonprofit.

“Philanthropy can catalyze investment in the foundational elements of public sector AI infrastructure, including skilling, talent, data, and computing capacity,” said Vilas Dhar, President, Patrick J. McGovern Foundation.

Examples from the field help ground the playbook in practice. Lessons from the IBM Impact Accelerator, IBM’s pro bono social innovation program, show how organizations can use technology to scale solutions to environmental and economic challenges. Those examples illustrate what responsible AI can look like across areas such as sustainable agriculture, water management and resilient cities. The playbook also incorporates insights from University College London’s Industry Exchange Network (IXN), which has connected thousands of students with more than 300 organizations since 2011 through a structured approach to real-world problem solving. Together, these examples and collaborations help make the guidance more practical and applicable across the social impact sector.

In addition to case studies and external insights, the guide points readers to resources that can support adoption, including IBM SkillsBuild for free AI courses and assessments, IBM’s Design Thinking Framework for human-centered development and tools such as watsonx.governance and Granite Guardian to help strengthen responsible AI practices.

Responsible Use of AI for Social Impact is available as a free download from NationSwell. The interactive report includes tailored navigation for different audiences, along with links to training, tools and frameworks that organizations can use to assess readiness and build capability over time.

Organizations can begin their AI journey immediately by taking the IBM SkillsBuild AI Level Up assessment to determine AI readiness and access curated learning plans.