How to Tell a Client You Used AI on Their Work

A client asks whether you used AI on the deliverable you just sent, or you decide to get ahead of the question before they ask it. Either way, the instinct to over-explain is strong, and it usually backfires. The client is not asking for a confession. They are asking two narrower questions: is this any good, and did anything sensitive leave my control.

Answer those two questions directly and the rest of the conversation takes care of itself.

Separate The Quality Question From The Confidentiality Question

Clients rarely care, in the abstract, whether a tool touched their document. They care whether you reviewed it, and whether their information went somewhere it shouldn’t have. Collapsing both into one nervous paragraph makes the client work harder to find the actual answer. Split them: one sentence on your review, one sentence on where their material went.

Lead With Your Judgment, Not With The Tool

The order matters more than the content. Open with what you did and confirmed, then mention AI as a detail of process, not as the headline. A client who reads “I used AI, so please note” before anything else has already started discounting the work before you get to the part that reassures them.

Say Exactly What AI Did And Did Not Do

Vague disclosure invites the client to imagine the worst version. “AI helped with this” could mean anything from a spellcheck pass to an unreviewed first draft. Name the actual scope, structuring, first-draft language, formatting, whichever it was, and state plainly what you personally checked.

Before:

Just wanted to flag that I used some AI tools while putting this together, hope that’s ok, let me know if you have any concerns about that.

After:

I used an AI tool to draft the initial structure of this report, then wrote and verified every figure and recommendation myself. Nothing containing your data was sent to a public AI service; the drafting tool runs locally on my machine.

The second version answers both real questions in two sentences and does not ask permission for something already done well.

Confidentiality Is The Real Question Underneath

Even when a client does not say it, confidentiality is what they are actually worried about. This is a live professional issue, not a hypothetical one. The American Bar Association’s 2024 guidance on generative AI tools recommends that lawyers get informed client consent before putting client information into a generative AI tool at all, precisely because many of these tools can retain or learn from what is typed into them. Read the ABA’s own framework if you want the reasoning behind it. The same caution applies outside law: if your client’s material is going into a public AI service by default, that is worth knowing before you send anything, not after.

This is also a fair moment to mention how you actually work, if it is true. Rewriting a client draft in a tool that runs the model locally, with nothing sent to a public service, is a meaningfully different claim from “I used ChatGPT,” and clients handling regulated or sensitive material will notice the difference.

See also is it safe to paste work emails into ChatGPT for the mechanics of what happens to pasted text, and how to tell colleagues you used AI to draft something for the internal version of this same conversation.

A Wrivio Context For A Reusable Client Disclosure Line

If you do client-facing work regularly, save this once instead of composing it from scratch under time pressure.

Rewrite this as a short, confident note to a client about AI assistance on their deliverable. Lead with what I personally reviewed and verified, then state specifically what AI helped with. Include one clear sentence about where their information was and was not sent. No apologizing, no asking permission for something already done. Keep every name, date, figure, and commitment exactly as written.

Press Ctrl+Shift+Space, paste your draft note, and check the diff against the original before sending, since this is a message where an accidental overclaim about verification is worse than saying nothing at all.

Common Questions

Should I disclose AI use before the client asks?

For deliverables the client relies on professionally, yes, a brief proactive note builds more trust than waiting to be asked and then explaining defensively.

What if the client’s material went into a public AI tool by mistake?

Tell them plainly what happened and what you are doing about it going forward; a fast, specific disclosure recovers trust far better than a vague reassurance discovered later.

Do I need my client’s consent before using AI on their work?

For sensitive or regulated material, get it in advance rather than assuming a general engagement letter covers it, since informed consent is now explicit guidance in some professions and good practice in all of them.

Is it enough to say “I used AI” without detail?

No, a bare statement invites the client to imagine the worst case; naming the specific scope and what you verified answers the question they actually have.

Does using a local AI tool change what I need to disclose?

It changes what you can honestly say about where the client’s data went, which is usually the part clients care about most, but you should still confirm what you personally reviewed.

Keep a client’s material off public AI services by working in a tool that runs locally: Download Wrivio for Windows.