Governance earlier than devices: the actual anxieties about AI in charity finance

Concern about AI is rising as quick as adoption. Here are the 5 worries value taking severely, and why none is a purpose to attend.

—————————————————————————————————–

The first time I pointed an AI device at a reconciliation I’d usually spend a morning on, the pace struck me as a lot as it could anybody. By the second run, the query that mattered wasn’t the way it had accomplished the work, however the place the information had been and who was accountable for the outcome.

That second query is the one I hear most from finance groups, and it deserves a correct reply. I spend my time now creating and deploying AI instruments for charity finance, and my present project spans a finance, services and digital remit. The nearer I get to the work, the extra satisfied I’m that the anxieties round it aren’t obstacles to adoption. They are the agenda for it.

Adoption is rising shortly. ACCA’s Global Talent Trends 2026 discovered that 52% of finance professionals now usually use AI, whereas 51% fear about its influence on jobs, up from 44% a yr earlier (ACCA Global Talent Trends 2026). And our nook is essentially the most cautious: charity and not-for-profit is essentially the most AI-recruitment-sceptical sector and joint-lowest on adoption, at 43%.

I do not learn that warning as an issue to speak individuals out of. I learn it as finance groups doing their job. The process is not to suppress the anxieties, however to work by way of them, so adoption occurs on a footing individuals can defend to their auditors, trustees and themselves. So let me take the concerns I hear most, within the order they have an inclination to reach.

Governance, and an precise coverage

Most charities begin utilizing AI earlier than they’ve written a line of coverage about it. A brief, plain AI coverage — what instruments are accredited, what knowledge could and should not go into them, who indicators off new makes use of — does extra to allow assured adoption than any know-how. It turns a nervous “are we allowed to?” into a transparent “yes, within these lines”.

Cyber and knowledge danger

Charities maintain a few of the most delicate private knowledge there may be, about donors, beneficiaries and employees. The self-discipline is easy to state and more durable to stay: delicate or private knowledge mustn’t go into instruments that are not accredited to carry it. Much of my very own AI work sits nicely away from private knowledge, on administration accounts and board reporting. Knowing the place the delicate materials is, and protecting it out of the flawed locations, is a lot of the battle.

Shadow AI

Whatever your coverage says, a few of your persons are already utilizing instruments you have not accredited, as a result of they’re stretched and these instruments assist. A ban does not cease that; it drives it underground, the place you possibly can’t govern it. Better to level individuals in the direction of accredited, registered instruments, make it simple to ask for a brand new one, and deal with curiosity as one thing to channel moderately than punish.

Training

It is straightforward to make use of these instruments badly. The distinction between sharper evaluation and confident-sounding nonsense is basically coaching. ACCA’s analysis is encouraging: 82% of finance professionals are assured they will study AI abilities, and 43% of employers now present upskilling, up from 32%. A modest, deliberate programme will return greater than the instruments themselves.

Cost

This is the most recent fear and the one I’d flag hardest. The extra succesful fashions are additionally dearer, and distributors are altering how they invoice because the market matures. Budget for AI intentionally, and watch the unit economics — what every use truly prices and saves — moderately than assuming at this time’s worth is mounted.

None of those 5 is a purpose to carry again. Each is a factor to get proper so you possibly can transfer ahead with out trying over your shoulder. The organisations that get essentially the most from AI aren’t the boldest, however the ones that do that groundwork first.

Which is why I discuss governance earlier than devices. The thrilling a part of this work is what it offers again to individuals: hours returned to the judgement and relationships solely an individual can supply. But you solely get pleasure from that if the foundations are sound. As ACCA places it, AI is a productiveness layer, not a substitute, and accountability doesn’t change. When the accounts are signed, the Finance Director nonetheless owns the numbers. Get the governance proper, and you do not have to decide on between warning and progress.

Alan de Sousa Caires FCCA is Director, Finance & AI Advisory, AI Finance Office

---