Should You Tell Customers You Use AI?

The short answer is: it depends on what the AI is doing. Not on how you feel about AI, and not on what your competitors are disclosing. The question is whether the customer has a reasonable expectation to know, and whether telling them builds or costs trust in your specific context.

This article lays out the real trade-offs so you can make a considered call. It is not legal guidance. If you operate in a regulated industry, such as financial services, healthcare, or law, you likely have specific disclosure obligations that sit outside this discussion entirely. Talk to your professional advisers about those.

Why this question is harder than it looks

A few years ago, the answer felt simple: disclose AI use and you look ahead of the curve. Now AI is embedded in spell checkers, customer service queues, email drafting tools, and analytics dashboards. Disclosure is no longer a single decision. It is a series of smaller ones, made differently depending on where AI sits in your process.

The useful frame is not "do we use AI?" It is "what is the AI doing, and does the customer have a stake in knowing?"

Where disclosure clearly helps you

When the customer is providing personal data

If a customer submits information and AI is processing it, summarising it, or making decisions based on it, tell them. This is not just about trust. Many customers will assume their data stays with a human unless you say otherwise. Correcting that assumption early is cleaner than correcting it after a complaint.

Examples where this applies:

• An AI tool that reads customer intake forms to generate a first-draft proposal
• A chatbot that stores conversation history to personalise future responses
• An AI system that scores leads or flags customer accounts

In these cases, say what the AI is doing with the data, and say it before the customer submits anything.

When the AI affects the final deliverable directly

If you are selling a written report, a content piece, a design, or any other deliverable, and AI has produced a significant part of it, the customer may reasonably feel misled if they find out later. Whether or not that is a legal issue in your jurisdiction is for your adviser to determine. The trust issue is separate and real.

"Significant" is the word that matters here. A copywriter who uses AI to generate a first draft and then rewrites it substantially is in a different position to one who edits lightly and sends it on. You will need to make a judgment about where your work sits.

When the relationship is built on personal expertise

Consultants, coaches, advisers, and specialists are often hired for their individual judgment. If a customer believes they are getting your thinking and they are mostly getting an AI output that you have reviewed, the gap between expectation and reality matters. Disclosure here is not about shame. It is about accuracy.

Where disclosure is expected and already baked in

Some AI use is now so standard that customers expect it even if they do not articulate that expectation.

• Automated email routing and tagging
• Spam filtering
• Grammar and tone suggestions in customer-facing writing tools
• Website personalisation
• Search and recommendation systems

Nobody expects you to disclose that your inbox uses AI to sort priority messages. The context makes the use obvious, the customer has no particular stake in it, and a disclosure would create noise rather than clarity.

The test here is whether a reasonable customer, knowing the AI was being used this way, would feel anything other than indifferent. If the honest answer is "they would not care," disclosure is unlikely to add value.

Where it is unnecessary and may even cause confusion

Disclosure is not always the safer option. In some cases, it introduces doubt about work quality where none is warranted, or it prompts questions that distract from the actual service.

Consider a small business that uses an AI tool to prepare cleaner invoices, track project time, or generate internal meeting notes. Telling every customer about this adds nothing to their understanding of the service. It may even prompt questions the customer did not have before.

The question is whether the AI use affects the customer's interests in any material way. If it does not, disclosure may be noise.

How data changes the calculation

The clearest signal that disclosure is warranted is customer data. Specifically:

• Is the customer's own data being passed to an AI model?
• Is that model external, meaning the data leaves your systems?
• Could the output of that AI process affect the customer in a meaningful way?

If the answer to any of those is yes, a simple, plain statement in your terms of service or your onboarding process is the minimum reasonable step. It does not need to be a warning. It needs to be a fact: "We use [AI tool] to [do this thing] with the information you provide."

This is not legal advice. It is a practical trust standard. Your privacy policy and your legal obligations are separate matters and may require more specific language depending on your jurisdiction and sector.

A plain test for deciding

Ask these four questions about each AI use case in your business:

1. Does it touch customer data? If yes, disclose what the AI does with it.
2. Does it directly shape the final deliverable the customer is paying for? If yes, consider whether the customer's expectation matches the reality.
3. Is it background infrastructure the customer has no stake in? If yes, disclosure is probably unnecessary.
4. Would a reasonable customer feel surprised or misled if they found out later? If yes, tell them now.

This is a trust test, not a legal checklist. Whether specific rules require disclosure in your context is a question for your legal or compliance adviser.

The regulated industry point, stated clearly

If your business operates in financial services, healthcare, legal services, insurance, or any other regulated sector, your disclosure obligations around AI may be set by regulation, not just by good judgment. Some regulators have published specific guidance on AI use in customer-facing processes. Others are still developing it.

Do not rely on this article to determine your compliance position. Get advice specific to your sector.

What good disclosure actually looks like

When disclosure is warranted, the goal is plain and factual. Not apologetic, not promotional.

Bad: "We are excited to harness the power of AI to deliver you a cutting-edge experience." Better: "We use AI to draft initial responses to your enquiries. A member of our team reviews each one before it is sent."

The second version tells the customer what is happening and who is responsible. That is the full job.

Putting it in writing

The right place to address AI disclosure formally is your AI usage policy. A clear policy documents which tools you use, what data they access, how outputs are reviewed, and what customers can expect. It gives your team a consistent position and gives customers a place to find answers without having to ask.

Plainstart offers a free AI Usage Policy template for small and medium businesses. It is written in plain language, covers the key areas most SMBs need to address, and is designed to be edited to fit your business. No signup wall.

If you want the full system, including a data governance checklist, a 90-day adoption plan, a tool scorecard, prompt libraries, and an ROI tracker, the complete AI Adoption Kit is available for $149.

AI adoption, done properly.