Should you tell customers when AI is involved?
Written lesson · about 3 minutes to read
The short answer
Yes, whenever a customer is talking directly to an AI, in chat or on the phone, tell them and give them an easy way to reach a person. A reply you have read and approved does not usually need a label, but you own every word, and trust is the business reason to be open.
Three things to remember
- 01If a customer is talking to an AI in chat or on the phone, tell them.
- 02Offer an easy route to a person, and never let an AI pose as a named human.
- 03A person-reviewed email drafted with AI usually needs no label, but you own every word.
The plain position
When a customer is talking directly to an AI, in your website chat or on the phone, tell them at the start. One line does it: 'You are chatting with our automated assistant, and a person is available whenever you ask.' Then make that true, with a clear way to reach a real person that does not make the customer argue with the machine first.
Some places already require this for automated conversations, and more rules are coming. What applies to you depends on where you and your customers are, and it can change. The sensible habit is to say it everywhere, whether or not a rule makes you.
When a label is not needed
A person-reviewed email is different. Suppose your office manager asks an assistant for a first draft of a reply, reads it, fixes what is wrong and sends it under her own name. That is a person writing with a tool, and most businesses do not label it, just as they do not label a spell-checker.
But you own every word. If the draft promised a date or a price, it is your promise. If it got a fact wrong, the mistake is yours. 'The AI wrote it' will not satisfy the customer. The person who presses send is the author, so read every draft as if you had written it.
What never to do
Never let an AI pretend to be a named human. A chat assistant called 'Sarah from the service team' with a stock photograph, or a phone voice that claims to be a real employee when asked, is a lie told to your own customers. When someone asks directly whether they are talking to a person or a machine, the answer has to be true.
Do not let it imply authority it does not have, either. It should never say a refund is approved when no person has approved it. A well set up assistant says it is automated, answers only from what you gave it, and hands over to a person for complaints, upset customers and anything it does not know.
What to do this week
Trust is the business reason to be open. A customer who finds out afterwards that they were talking to a machine may stop trusting everything else you told them. Automation that is quick, honest and easy to escape gives customers little to complain about.
Read your website chat greeting and your phone greeting as a customer would. Check that the first line says it is automated and that a person is easy to reach, and fix both if not. Then check the rules where you operate, with a lawyer for your own case, because requirements differ by place and are changing.
More in part 3
- 1AI security for small business
- 2Shadow AI: your team is already using it
- 3Where does your data actually go?
- 4Is your data being used to train their model?
- 5Personal or business AI accounts: why it mattersWritten
- 6What is PII, and how do you keep it out of AI?
- 7What is data retention, and who decides it?
- 8How are scammers using AI, and how do you stop them?Written
- 9What is prompt injection?
- 10How to write an AI policy for a 20-person company
- 11Recording calls with AI: what do you have to tell people?Written
- 12Should you tell customers when AI is involved?WrittenYou are here
- 13Who owns what AI creates?
- 14AI and your insurance: who's liable when it's wrong?
Next step
Where does this fit your business?
Six quick questions about your business, then your AI plan: what to start with, what a person still checks, and the lessons that go with it. Or bring it straight to a free thirty-minute call.