There's a huge difference between "this company uses AI" and "this company deceived me about how it uses AI," and that difference is exactly where customer trust is won or lost. The vast majority of people have no problem with a business using artificial intelligence to handle enquiries, generate content or automate processes. The problem arises when the customer discovers, after the fact, that they were actively kept from knowing they were interacting with an automated system, or that a decision affecting them (a rejected complaint, a price different from another customer's) was made by an algorithm with nobody telling them or offering them a human review.
Why transparency isn't just an ethical issue, it's a legal one
The European Union's AI Act, being phased in progressively, sets out specific transparency obligations for certain public-facing uses of artificial intelligence. Among them, the obligation to clearly inform a person when they're interacting with an AI system (for example, a chatbot) rather than a human, unless that's obvious from context. GDPR already established, before the AI Act, the right of any person not to be subject to a decision based solely on automated processing that produces legal effects or significantly affects them, without being able to request human intervention. Failing to meet these obligations isn't just a reputational risk, it's a real legal and enforcement risk.
The moments where transparency matters most
- Chatbot customer service. The customer should know, clearly and without having to ask, that they're talking to an automated system, especially at the start of the conversation, and should always have a simple way to ask for a human.
- Automated decisions affecting the customer. Rejecting a credit application, denying a complaint, calculating a personalised price: when AI plays a decisive role in a decision like this, the customer has the right to know and to request a human review.
- Content generated or altered with AI. Images retouched or generated with AI in advertising (especially if they substantially alter the real appearance of a product or person), or automatically generated marketing content presented as if it were a genuine human testimonial.
- Using personal data to train or personalise AI systems. If a customer's data is used to train models or personalise experiences beyond the basic expected use, it must be clearly disclosed, not buried in the small print of a twenty-page privacy policy.
A textbook case of what not to do
There have been widely documented cases of companies across different sectors rolling out chatbots with human names and realistic profile photos, with no indication whatsoever that it was an automated system, sparking considerable backlash when users discovered it on their own. The reputational damage in these cases didn't come from using AI itself (which most users would have happily accepted if communicated clearly), but from the feeling of having been deliberately deceived about something the company knew mattered.
Why hiding it is usually worse business than disclosing it
There's a temptation to think hiding AI use makes the experience feel "more human" and therefore more satisfying for the customer, but that logic ignores a decisive factor: almost everything that's hidden eventually comes out, and the reputational cost of being caught hiding something is systematically higher than the cost of having disclosed it from the start. Companies that openly disclose AI use don't just comply better with regulation, they also protect themselves against that asymmetric risk: the worst that can happen if you disclose it is someone being indifferent; the worst that can happen if you hide it is an avoidable trust scandal.
How to communicate AI use without sounding like a cold legal disclaimer
Transparency doesn't need to feel like a sterile legal notice bolted on awkwardly. A simple "Hi, I'm [brand]'s virtual assistant, here to help with the basics, and if you'd rather talk to a person just ask" fulfils the transparency obligation and, well worded, doesn't drain any warmth from the experience. The key is integrating transparency into the brand's normal tone, not treating it as a separate legal formality tucked away somewhere barely visible.
The balance between efficiency and humanity
Being transparent about AI use doesn't mean apologising for using it, or giving up the efficiency it brings. It means giving the customer the information and control they need to feel treated with respect: knowing who or what they're talking to, and always having the option to escalate to a person when they need it. The businesses that handle this transition best aren't the ones avoiding AI out of fear of customer backlash, they're the ones using it with good judgement and being honest about how they use it.
A positive case: the insurer that won trust by being transparent
A mid-sized insurer introduced an AI system for the first stage of processing simple claims, openly communicating on its website and within the process itself that this first assessment was done by an automated system, with human review guaranteed in any case of doubt or customer disagreement with the outcome. Far from generating pushback, the clarity about the process (combined with the explicit guarantee of available human review) created a perception of honesty that, according to their own satisfaction surveys, improved overall trust in the brand, even among customers who never actually needed that review.
Training the internal team on transparency, not just writing the notice
A well-written transparency notice doesn't help much if the rest of the team (customer service, sales, marketing) doesn't understand or can't naturally explain how and where AI is used in the business. When a customer directly asks "did a person decide this, or a computer?", the answer should be consistent and clear regardless of who inside the company gives it, not depend on each employee improvising in the moment. Including brief training on this topic in the onboarding of anyone dealing with customers avoids contradictory answers that, paradoxically, generate more distrust than the original question itself.
Common mistakes when communicating AI use to customers
The first frequent mistake is limiting transparency to a single point in the customer journey (for example, a notice at the start of a chat) and forgetting the same obligation can apply at other moments where AI is also involved without anyone having considered it, like an automatically generated follow-up email or a product recommendation based on a personalisation model. Transparency isn't a formality ticked off once, it's a principle that needs applying consistently at every point where AI plays a meaningful role in the customer experience.
The second mistake is writing the transparency notice in language so technical or legalistic that, while it meets the letter of the rule, in practice it communicates nothing understandable to the average customer. A notice nobody understands poorly serves its real purpose, which is for the customer to genuinely know who or what they're interacting with, not just for some text to technically exist somewhere mentioning it.
The third, deeper mistake is assuming transparency is exclusively a legal or compliance department problem, without involving whoever designs the customer experience. Transparency communicated with good tone and naturally woven into the conversation flow works far better, and creates less friction, when it's designed by someone who understands user experience, not just someone drafting legal clauses. The businesses that handle this best treat transparency as part of product design from the start, not as a legal patch bolted on at the end of the process.
The fourth mistake is not checking how different customer segments perceive transparency, assuming a single message works equally well for everyone. A customer used to interacting with automated assistants across many other services may barely register the notice, while another with less technological familiarity may need a somewhat more detailed explanation to feel comfortable. Adjusting the level of detail based on channel and typical customer type, while still meeting the basic obligation to clearly inform, improves the experience without sacrificing regulatory compliance.
The fifth, more strategic mistake is treating transparency as a cost to minimise rather than an opportunity for differentiation. In a market where more and more businesses use customer-facing AI, disclosing it honestly and in good tone can become a trust argument against competitors who hide it or communicate it poorly, rather than simply a regulatory requirement to meet with the least possible effort. Brands that understand this before the rest of the sector usually build a trust advantage that's hard for latecomers to quickly replicate.
The sixth mistake is not distinguishing between what the law requires and what a brand voluntarily goes beyond, presenting both as if they were the same thing. Being transparent about what regulation requires is mandatory; going further and voluntarily explaining in detail how a system works, even when not strictly necessary, is a brand decision worth making consciously, not confused with mere minimum regulatory compliance.
Periodically reviewing which AI uses have been added without disclosure
Over time, it's common for a business to incrementally add new AI uses (a new chatbot, a lead-scoring system, a price personalisation tool) without each addition going through an explicit review of what transparency obligations it carries. A periodic audit, even a simple one, of every point where AI interacts with the customer or makes decisions affecting them, helps catch communication gaps that have built up with no bad intent, simply because each tool was implemented separately without an overall view.
Frequently asked questions
Am I legally required to tell customers they're talking to a chatbot and not a person?
Yes, in general the European regulatory framework requires informing people when they're interacting with an AI system rather than a human, unless it's obvious from context. It's worth checking the exact requirements applicable to your sector with legal advice, since the level of obligation varies by type of use.
What happens if my AI system makes a wrong decision about a customer?
In most regulated cases, the customer has the right to request a human review of an automated decision that significantly affects them. It's important to have a clear, accessible process for that review, not just comply with the transparency notice.
Do I have to disclose if an advertising image was generated or retouched with AI?
It depends on the context and how far the image strays from the reality of the product or service; if the alteration could mislead about the real characteristics of what's being sold, it should be clearly disclosed, both for advertising ethics and possible legal risk of misleading advertising.
Does transparency about AI scare off customers or generate distrust?
The evidence suggests the opposite when communicated well: what generates distrust is discovering undisclosed AI use, not having it communicated normally and in good tone from the start.
How do I integrate AI transparency without it sounding like a heavy legal notice?
Write it in the same warm tone you'd use for any other brand communication, integrated naturally into the first point of contact (for example, at the start of a chatbot conversation), rather than as a separate, cold block of legal text.
Do these obligations also apply to small businesses, or only to large corporations?
They apply regardless of company size; European regulation doesn't exempt small businesses from transparency obligations around customer-facing AI use, although the complexity of compliance is usually proportional to the type of use involved.