Fewer than 1 in 4 UK charities have approved AI policies, report warns

TL;DR: A new Charity Excellence Framework report finds AI adoption across UK charities running well ahead of governance, with fewer than one in four having approved tools, written policies and training in place. The risk, the report argues, is a loss of public trust — particularly around AI-generated imagery in fundraising. Trustees remain legally responsible for AI risks even when boards do not know the tools are in use.

The findings echo a March 2026 University of East Anglia study, which warned charities could break a “bond of trust” with the public through AI imagery in campaigns. The Charity Excellence Framework’s framing is sharper still: only a very small minority of UK charities are deploying AI with clear organisational oversight, agreed policies or trustee-level ownership, even as use accelerates.

Where the gap is widest

AI-generated imagery is the specific pressure point. Many UK charities have piloted it and pulled back, citing ethical, reputational or authenticity concerns. Public sentiment, the report notes, is conditional rather than hostile — strong support for authentic photography, lower acceptance for AI imagery that appears realistic or emotionally manipulative, and particular sensitivity in contexts involving vulnerable people.

The governance gap is structural rather than philosophical. Most UK charities are early in their AI journey, with staff and volunteers using tools such as Microsoft Copilot largely informally. Trustee-level visibility into what tools are deployed, on what data, and with what consent is patchy. Yet under charity law, trustees retain legal accountability for AI-related risks including data protection, safeguarding and bias — risks that do not disappear because the board has not yet asked about them.

Looking forward

For UK third-sector leadership, the report points to three pragmatic moves: a strategic AI assessment commissioned at trustee level, named committee or trustee responsibility for AI use, and organisation-wide training and compliance. The deeper editorial point is that charities sit in a different trust position to commercial users: regulators, donors and beneficiaries already scrutinise integrity in ways that for-profit AI deployments rarely face. Charities that move first to publish AI use disclosures — what tools, on what data, with what human oversight — are likely to convert governance discipline into a trust advantage. Those that wait until a fundraising image is flagged in the press will pay a higher reputational price than the deployment ever saved.