Customer communication
Should you tell customers you use AI?
Nobody expects a disclaimer on a document because you used a spellchecker. Somewhere between that and a chatbot pretending to be a person, disclosure starts to matter — and it is worth knowing where the line sits before a customer finds it for you.
Sometimes it is not your choice
Start with the part that is not a judgement call. In a growing number of places you are legally required to make clear when someone is interacting with an automated system rather than a person, and the direction of travel is towards more of this rather than less.
If you run a chatbot, an automated phone system, or anything a customer could mistake for a human conversation, say so at the start. That is not a marketing decision, and "they probably worked it out" is not a defence.
Rules vary by country and by sector, and they change. If a meaningful part of your customer contact is automated, this is worth twenty minutes with someone who knows your jurisdiction rather than a guess.
The useful test
Outside the legal cases, one question resolves most of it: would this customer feel misled if they found out?
Not "would they mind" — people mind all sorts of things. Misled. Did they believe something about the interaction that was not true, and did that belief matter to them?
| Situation | Would they feel misled? | Disclose? |
|---|---|---|
| You drafted a product description with AI and edited it | No | No |
| AI summarised your meeting notes | No | No |
| A chatbot answers questions on your site | Yes, if it seems human | Yes, up front |
| A personal apology was AI-written and unedited | Yes | Do not send it at all |
| AI helped analyse their data to make a recommendation | Possibly | Say so when you advise |
The pattern: tools that help you produce work do not need announcing. Anything that replaces the human relationship a customer thought they had does.
The case where silence costs you
The expensive scenario is not being caught using AI. It is sending something personal that was obviously not personal.
A customer writes a real complaint — something went wrong, they are upset, they took time to explain. They get back a fluent, structured, faintly warm reply that addresses none of the specifics. They know immediately. And what they learn is not "this business uses AI", it is "this business did not read my email".
That is the damage. It is not about the technology at all, and no disclosure fixes it. The fix is to write that one yourself.
When saying so actually helps
There are cases where disclosure is an advantage rather than a risk.
- It explains speed. "Our system drafts a first response immediately; a person reviews it before it reaches you." That is reassuring and true.
- It sets expectations on a bot. Customers are far more patient with an automated assistant they were told about than with one they discover.
- It is part of your positioning. If you sell efficiency, using modern tools is evidence rather than a confession.
- Your customers are technical. They assume it already, and pretending otherwise looks naive rather than careful.
How to say it without a paragraph of hedging
Disclosure does not need to be a policy statement. One plain sentence, in the place it matters.
- On a chatbot: "You are chatting with our automated assistant. Type agent at any time for a person."
- On a fast reply: "This first response was drafted automatically and checked by our team."
- On a recommendation: "We used automated analysis to produce this — happy to walk you through the reasoning."
Short, specific, no apology. A long disclaimer suggests you think you have done something that needs excusing, which invites the customer to agree.
The line worth holding
Be honest about the relationship, not exhaustive about the tooling. Customers do not need an inventory of your software. They do need to know whether they are talking to a person, and to trust that when something matters, someone read it.
Get that right and the question of whether to disclose mostly stops being difficult.