Guardrails for accountants and bookkeepers

When you finish, you can decide what never goes into an AI tool, how to verify output, and how to follow professional rules.

Stuck? Ask the AI tutor

Client financial data is confidential

Clients trust you with bank details, payroll, tax numbers, personal income and business secrets. Professional bodies and privacy laws in most countries expect you to keep that confidential. Many firms also have contractual duties to clients.

Do not paste client names, tax or ID numbers, bank account or card numbers, payroll details, or complete ledgers into a consumer AI tool unless your firm has approved that tool for that data. Use made-up numbers and generic labels such as [Client A] and [Supplier 1] when you ask for help.

Accuracy: numbers, rules and invented sources

AI tools can do arithmetic wrong, mix up tax rules from different countries, quote a rule that has changed, or invent a citation that looks real. A nice-looking paragraph can hide a false statement. Treat all numbers, rule claims and references as unverified until you check an official source.

A good habit is to ask the tool to show its reasoning step by step, then re-calculate in your normal software. If it names a standard or a law, look it up yourself. If you cannot find it, do not use it.

Your professional duty and no advice from a draft

If you hold a licence or belong to a professional body, its code of conduct applies when you use AI. You remain responsible for your work. Using a tool does not shift blame. Do not give clients tax, legal or investment advice generated by a tool without your own professional review and the right qualification.

If you are a bookkeeper who does not give tax advice, stay inside that boundary. A tool may confidently suggest a tax approach. Your job is to say, “That is a question for a qualified tax adviser,” rather than pass a draft along.

Bias, disclosure and saying no

Watch for bias in anything that scores or flags people, such as rating a client’s risk from a description. Tools can reflect stereotypes. Do not use AI alone to decide that a client or a vendor looks suspicious.

Follow your firm’s policy on telling clients when AI is used. Where the policy is silent, being open is a good default for anything that shapes a deliverable. And when a colleague or client wants a shortcut that puts data at risk, you can say no and suggest an approved route. Protecting client data is part of the work.

Practice now

Write a short list of data types in your work that must never go into a consumer AI tool, then check it against your firm's policy or ask your manager what the policy is.
Ask a free AI tool a rule question about your country's reporting requirements. Then look up the answer on an official site and note where the tool was wrong or out of date.

Use any AI assistant you have. Do not paste private, client or patient information into it.

Check yourself

1. A colleague pastes a client's full ledger with names into a free chatbot to get a summary. What is the main problem?

2. An AI tool cites a regulation by number to support a treatment. What should you do?

3. A bookkeeper who does not give tax advice gets an AI suggestion on tax strategy. What should they do?

Ask the AI tutor

Ask anything about this lesson, in any language. The tutor is an AI: it can be wrong, and it never needs your personal details.

Next lesson: Grow with AI: your 30-day plan

AI-drafted. Checked by an automated validator and a separate AI reviewer; a person has not read every lesson. Translations and tutor answers are AI-made. Found a mistake? Tell us. AI can be wrong. General information only, not professional advice. Not affiliated with any AI company named. Current University is not accredited.