AI Literacy for Charity Trustees and Boards: A Plain-English Briefing
Twenty-eight per cent of charity boards admit poor digital skills. Most have had no briefing on AI. With Article 4 enforcement weeks away, this is the plain-English briefing your trustees should have had a year ago.
Twenty-eight per cent of charity boards admit they have poor digital skills. Most have had no formal briefing on AI at all. Article 4 enforcement begins in twelve weeks. This is the briefing your trustees should have had last year.
The governance gap nobody is talking about
The 9th UK Charity Digital Skills Report 2025 found that twenty-eight per cent of charity boards rate their digital skills as poor. Thirty-six per cent of charity CEOs say the same about themselves. Most boards have never had a structured conversation about AI use in their charity.
This is the same group of people legally responsible for the prudent management of the charity's affairs.
The Charity Commission has not issued AI-specific guidance for trustees, but its existing guidance on duty of prudent management, mission alignment, and risk oversight already applies. The EU AI Act's Article 4 enforcement deadline of 2 August 2026 sits on top of that. Trustees are accountable for AI use in their charity from now on, whether they know it or not.
This article is the briefing trustees need. It is written for trustees to read directly. CEOs can pass it to their boards. Trustees can use it to brief themselves.
It is not training. It is the structured briefing that should precede training.
What trustees actually need to know about AI
The popular framing of AI for charity boards is technical. How large language models work. What prompting is. What a context window is. Most of this is irrelevant to the trustee role.
Trustees do not need to know how AI works. They need to know what AI is doing in their charity, what could go wrong, and what controls exist.
The questions below are what trustees should be able to answer about their charity's AI use. If your board cannot, you have a governance gap that needs closing.
1. What AI is being used in our charity, and for what?
Most boards have no inventory. AI tools enter charities bottom-up. A fundraiser uses ChatGPT. A communications officer uses Copilot. A service manager uses an AI summary tool. None of it has been approved at trustee level because nobody asked.
Trustees should know:
- The list of AI tools in use, official and unofficial
- What each is used for
- Who in the charity uses which
- What data each handles
This is not granular knowledge. It is the kind of summary that fits on a single page. If the CEO cannot produce that page, the board's first AI agenda item should be commissioning it.
2. What policy governs our AI use?
If the charity has an AI policy, trustees should have read it. If the charity does not, that is also worth knowing.
Trustees should know:
- Whether a written AI policy exists
- What it says about approved tools, data handling, and accountability
- When it was last reviewed
- Whether it has been formally approved by the board
A policy that has not been seen or approved by trustees is not a charity policy. It is an operational note. Article 4 enforcement and Charity Commission expectations both treat trustee approval as the relevant standard.
3. Where is our AI use highest risk?
Not all AI use carries the same exposure. Trustees should be able to identify the two or three areas of the charity where AI use is highest stakes. That is usually:
- Anywhere beneficiary data is involved
- Anywhere safeguarding considerations apply
- Anywhere AI is informing decisions about people (admissions, eligibility, prioritisation)
- Anywhere AI-generated content reaches public audiences in the charity's name
- Anywhere AI is being used in fundraising appeals or grant applications
Trustees do not need to know the operational detail. They do need to know the risk map.
4. Does the EU AI Act apply to us?
Most UK charity trustees have not been told that the EU AI Act applies to their charity. Most assume it does not. Most are wrong.
The Act applies to UK charities with any EU exposure: staff, beneficiaries, partners, funders, audiences, or service delivery touching the EU. Most UK charities have at least one of these. Our guide on [whether the EU AI Act applies to UK charities] walks through the eight-question test.
Trustees should know whether the Act applies to their charity, and if so, what obligations it creates. The two most likely to apply are:
- Article 4 (AI literacy), enforced from 2 August 2026
- Article 50 (transparency obligations), enforced from 2 August 2026
Both have financial and reputational exposure for non-compliance.
5. What AI literacy training has our team had?
This is where most boards realise the gap. Article 4 requires documented evidence of AI literacy training. "Our team plays with ChatGPT" is not evidence.
Trustees should know:
- What AI training staff have received
- Whether it was role-specific or generic (only role-specific meets Article 4's "sufficient level" standard)
- What evidence exists (attendance records, content, assessment)
- What the plan is for August 2026
If the answer is "nothing structured," the board has a decision to make about resourcing and timeline. See our guide on [why generic AI training fails charities] for what role-specific actually looks like.
6. How would we know if AI use went wrong in our charity?
This is the question that surfaces the governance gap most clearly. Most boards have no mechanism to know.
Trustees should know:
- Whether staff are reporting AI mistakes or near-misses
- Whether trustees are briefed on AI use as part of routine reporting
- What the escalation pathway is if a serious AI-related incident occurs
- Whether the charity has any insurance considerations specific to AI
A charity with no reporting mechanism for AI incidents has no early warning system. The first sign of a problem will be the problem itself.
7. What is our position on AI ethics?
This is the question trustees are best placed to answer and most often skip. Operational staff make AI decisions inside an ethical frame the charity has not articulated. Trustees own the frame.
Trustees should be able to answer:
- Are there AI uses that conflict with our values, even if they would save us time?
- How would our beneficiaries feel if they knew we used AI in specific parts of our work?
- How would our donors feel?
- Are there areas of our work where we will not use AI at all, on principle?
These questions cannot be answered by an operations team. They are the trustees' work.
The 4Ps Framework as a trustee tool
The [4Ps AI Literacy Framework] (Purpose, Policies, People, Practice) is GoodAgents' methodology for charity AI literacy. For trustees, it is also a useful diagnostic.
Run the charity through each P from the board's perspective:
- Purpose. Has the board articulated where AI use serves the charity's mission and where it does not?
- Policies. Has the board approved a current AI policy?
- People. Has the board been briefed, and does it know what training staff have received?
- Practice. Does the board see evidence that AI use is being managed well, not just enabled?
A board that cannot answer yes to all four is not yet meeting its oversight responsibility. A board that can answer yes is in the front quarter of the UK charity sector.
The trustee briefing structure that works
We have run AI briefings for trustee boards across the UK charity sector. The format that lands consistently is ninety minutes, structured as follows.
Ten minutes: context and regulatory landscape. What the EU AI Act is, what it requires, why it applies to this charity. Article 4 enforcement timeline.
Twenty minutes: current AI use in the charity. What tools are in use. What data they handle. Where the risk concentrates. Done in advance through a quick audit, presented to the board.
Twenty minutes: the 4Ps diagnostic. Where the charity sits on each P. What good would look like. What gaps exist now.
Twenty minutes: trustee-specific decisions. Policy approval. Risk appetite. Ethical positions. Resourcing for training. Reporting cadence.
Twenty minutes: questions and next steps. Documented decisions. Named responsibilities. Review date.
Ninety minutes. Once. Then quarterly updates as a fifteen-minute standing item on the board agenda.
This is not technical. It is the kind of governance work charity boards apply routinely to safeguarding, finance, and data protection. AI is the next item.
What this is not
This briefing is not training in the Article 4 sense. Trustees attending a briefing do not satisfy the charity's AI literacy obligation for trustees, although it is a useful first step.
Trustees who actively oversee AI use, set the charity's position, approve the policy, and engage with AI as a standing governance matter are demonstrating the kind of literacy the Act expects of governance bodies. The briefing is the starting point, not the end.
For the operational team, AI literacy is role-specific training in tools and decisions they will use day-to-day. For trustees, AI literacy is strategic literacy in oversight, risk, and ethics. The two are not the same and should not be delivered in the same format.
Three things to do this week
If you are a trustee reading this:
Ask your CEO for the AI inventory. If one does not exist, ask for one to be produced. This is the minimum the board needs to discharge its oversight responsibility.
Put AI on the next board agenda as a substantive item. Not "AI update" as a closer. A proper thirty-minute discussion with documents and decisions.
Read the [EU AI Act for UK Charities] guide. The pillar is written for charity leaders. It is the regulatory context your board needs.
If you are a CEO reading this and your board has not had an AI briefing:
Commission one before the next board meeting. Internal if you have the capability. External if you do not.
Use this article as the pre-read. Send it ahead of the meeting. Ask trustees to come ready to answer the seven questions for the charity they govern.
Document the outcome. What was decided. What the trustee position is on each P. What the next review will look at.
A ninety-minute trustee briefing now is the difference between a board that can answer regulator and funder questions in 2026 and one that cannot.
The honest read
Most UK charity boards are behind on AI. Most do not know how far behind. The August 2026 deadline closes that gap whether trustees are ready or not.
The work to close it is not difficult. It is structured, finite, and within the capability of any board willing to engage. The cost of not closing it is asymmetric: minor inconvenience if you do, real exposure if you do not.
If you would like help running a trustee AI briefing in your charity, [book a call]. We deliver these regularly. The structure works. The work matters.
This article is informational and does not constitute legal advice. Trustees of UK charities with specific AI compliance questions should obtain qualified legal advice in addition to operational guidance. Charity Commission guidance on prudent management, mission alignment, and risk oversight applies to AI decisions made by trustees.
Similar Posts
Forget prompting tutorials. Every person on your charity payroll needs to understand four things about AI before they use it for anything. The plain-language briefing for staff who never had the training.
Two-hour ChatGPT masterclasses are the default in the charity sector — and the most common reason AI literacy fails to land. Why generic doesn't work and what role-specific training looks like.
When EU AI Act enforcement begins, regulators will ask how you decide which AI tools staff can use. A working tool approval process — and the evidence trail it produces — is your compliance defence.