An AI transparency statement for customer service must include, at a minimum: which AI systems are used, what they are used for, whether customers are speaking with an AI or a human, what data is processed, and how customers can object or reach a human representative. This information is not only a matter of good service but also a legal requirement under the EU AI Act. In this article, we answer the most frequently asked questions about exactly what an AI transparency statement entails and how you can practically draft one for your customer service environment. Would you like to learn more about how AI can be used in customer service? Read on to learn more about AI-driven intelligence.
What is the minimum information that an AI transparency statement must include?
An AI transparency statement must, at a minimum, describe which AI systems you use, the purpose of each system, whether and when customers interact with AI, what personal data is processed, how long that data is retained, and how customers can contact a human representative or object to automated decisions.
In practical terms, this means that for every AI system in your customer service department, you should define the following elements:
- System Identification: What is the system, who developed it, and what role does your organization play (as a user or provider)?
- Purpose and Function: What is the system used for—for example, routing calls, answering frequently asked questions, or sentiment analysis?
- Notice to the customer: When and how is the customer informed that he or she is communicating with an AI system?
- Data Processing: What personal data is processed, on what legal basis, and how long is it retained?
- Customer Rights: How can a customer speak with a human representative, appeal a decision, or request an explanation of an automated outcome?
A transparency statement that simply states in general terms, “We use AI,” is insufficient. Customers need to understand exactly what happens during their interactions and how AI affects the outcome of their inquiry or complaint.
What does the EU AI Act say about transparency in customer service?
The EU AI Act requires organizations that use AI in customer service to actively inform customers when they are communicating with an AI system. This applies specifically to systems designed to interact with people, such as chatbots and voice assistants. The disclosure that you are speaking with an automated system must be clear and understandable before or at the start of the conversation.
The AI Act distinguishes between different risk levels. Most customer service AI falls into the “limited risk” category, which is subject to less stringent transparency requirements. However, systems that make decisions with significant consequences for customers—such as denying a request or prioritizing a complaint—may be classified as “high risk.” Stricter requirements apply to those systems, including the customer’s right, under Article 86, to request an explanation of the factors that determined a decision.
Practical implications for 2026: The compliance requirements for high-risk systems covered by Annex III will become enforceable on August 2, 2026. Organizations that use AI—for example, to prioritize customer inquiries or assess customer profiles—would be wise to start keeping a register of all AI systems now and to define their roles within those systems. The AI literacy requirement—which mandates that employees understand how the AI systems they use work—has been in effect since February 2025.
How does an AI transparency statement differ from a privacy statement?
An AI transparency statement and a privacy statement overlap to some extent, but serve fundamentally different purposes. A privacy statement explains how you handle personal data in accordance with the GDPR. An AI transparency statement explains how your AI systems work, what role they play in customer interactions, and what rights customers have regarding automated decisions.
The difference lies in the focus:
- Privacy Statement: This covers data. What data do you collect, why, on what legal basis, how long do you retain it, and who should you contact to request access or deletion?
- AI Transparency Statement: Covers systems and decisions. Which AI systems do you use, for what purposes, how do they affect the customer experience, and how can a customer challenge an automated decision or request human assistance?
In practice, it is advisable to combine both documents or to explicitly cross-reference them. When an AI system processes personal data—which is almost always the case in customer service—both statements apply. A DPIA (Data Protection Impact Assessment) may be required when AI systems process personal data on a large scale or engage in profiling.
How do you explain the use of AI without technical jargon?
You explain how AI is used without technical jargon by focusing on the customer’s experience, not on how the system works. Describe in plain language what the AI does during the interaction, why that’s helpful for the customer, and what the customer can do if he or she would prefer to speak with a human.
A few practical guidelines:
- Use the term “digital assistant” or “automated system” instead of terms such as “large language model” or “NLP engine.”
- Explain what the system does for the customer: “Our digital assistant answers frequently asked questions right away, so you don’t have to wait.”
- Be honest about limitations: “For complex questions, you’ll be transferred to a representative.”
- Always provide a clear way out: “You can press ‘0’ or say ‘representative’ at any time to speak with a person.”
- Avoid double negatives and long sentences. Write as if you were explaining it to a customer over the phone.
The goal of clear language is not only compliance, but also trust. Customers who understand what is happening are less likely to be suspicious and view the AI interaction as a form of good service rather than an obstacle.
Where should an AI transparency statement be visible to customers?
An AI transparency statement must be visible at the moment the customer interacts with an AI system. This means: at the start of a chat conversation, in the IVR greeting during a phone call, on the webpage where a chatbot is active, and as a separate section in your privacy policy or terms of service.
Visibility has two dimensions: active reporting and easily accessible documentation.
Active disclosure means informing the customer at the time of contact. A chatbot that does not identify itself as AI does not meet the transparency requirements of the EU AI Act. A simple opening line such as “You are speaking with a digital assistant from [organization name]” is already a good first step.
“Accessible documentation” means that the full AI transparency statement is available on the website, preferably via a direct link in the chat interface or in the confirmation email sent after a customer interaction. Customers who want to know more about how AI affects their interaction should be able to find this information without having to search for it.
For organizations that use multiple channels—such as phone, chat, WhatsApp, and email—it is advisable to specify which AI systems are active on each channel. This prevents confusion and makes the statement practically useful.
When should an AI transparency statement be updated?
An AI transparency statement must be updated as soon as there is a change in the AI systems you use, the way they are used, the data they process, or the rights that customers have. Examples include introducing a new AI system, expanding an existing system to a new channel, or changes in legislation.
Specific situations in which an update is required or strongly recommended:
- You are introducing a new AI system into your customer service process.
- An existing system is being enhanced with new features that affect customer interactions.
- You switch to a different supplier for an AI component.
- Legislation changes, such as the entry into force of new obligations under the EU AI Act on August 2, 2026.
- An incident has occurred in which AI made an incorrect or harmful decision.
It is advisable to review the AI transparency statement at least once a year, even if there is no immediate reason to do so. Record the date of the last update in the document itself so that customers and regulators can see that the statement is up to date. Organizations that maintain an internal AI registry can link the transparency statement to that registry for a more efficient update process.
How Pegamento Helps Ensure AI Transparency in Customer Service
At Pegamento, we understand that AI transparency is not only a legal obligation but also an opportunity to strengthen customer trust. We help organizations use AI in a responsible and understandable way within their customer service environment. Our approach is concrete:
- Understanding Your AI Landscape: We help you create an overview of all AI systems in use in your customer interactions, including their roles and risk classifications.
- Transparency Built In: Our Agentic AI for customer service is designed with clear customer communication as the standard, not an afterthought. Customers always know when they’re interacting with a digital assistant and how to reach a human agent.
- Everything under one roof: From AI implementation to documentation and management, you have a single point of contact for the complete package. No silos, no complex supplier management.
- Customized solutions using standard building blocks: We combine proven modules to create a solution that fits your organization, industry, and customer contact volume—without the need for costly customization.
- Compliance-focused: We are ISO 27001 (information security), ISO 9001, and ISO 26000 certified, and we actively monitor developments related to the EU AI Act.
Would you like to know how your organization can use AI transparently and effectively in customer service? Contact us, and we’d be happy to help you explore your options.
Frequently Asked Questions
Geldt de AI-transparantieverklaring ook voor kleine organisaties met slechts één chatbot?
Ja, de transparantieverplichtingen onder de EU AI Act gelden ongeacht de omvang van je organisatie of het aantal AI-systemen dat je inzet. Zelfs als je alleen een eenvoudige FAQ-chatbot gebruikt, ben je verplicht klanten actief te informeren dat ze met een geautomatiseerd systeem communiceren. Voor kleine organisaties is de praktische last gelukkig beperkt: een korte openingszin in de chatinterface en een paragraaf in je privacybeleid kunnen al volstaan voor systemen met een laag risicoprofiel.
Wat is de grootste fout die organisaties maken bij het opstellen van een AI-transparantieverklaring?
De meest voorkomende fout is een verklaring opstellen die te generiek is, zoals ‘wij maken gebruik van AI-technologie om onze dienstverlening te verbeteren’, zonder te specificeren welke systemen er actief zijn, waarvoor ze worden ingezet en welke rechten de klant heeft. Zo’n vage formulering voldoet niet aan de eisen van de EU AI Act en wekt eerder wantrouwen dan vertrouwen. Een goede transparantieverklaring is systeem-specifiek, begrijpelijk voor een gemiddelde klant en bevat altijd een concrete route om menselijk contact te krijgen.
Hoe ga ik om met AI-systemen van externe leveranciers in mijn transparantieverklaring?
Als je AI-systemen van externe leveranciers inzet, blijf jij als organisatie verantwoordelijk voor de transparantie richting je klanten. Je moet in je verklaring vermelden welke derde partijen AI-componenten leveren en welke rol zij spelen in de verwerking van klantgegevens. Leg in je leverancierscontracten vast welke documentatie de leverancier moet aanleveren, zodat je je eigen transparantieverklaring accuraat en actueel kunt houden. Controleer ook of de leverancier zelf voldoet aan de verplichtingen die de EU AI Act aan aanbieders stelt.
Moeten medewerkers in de klantenservice op de hoogte zijn van de inhoud van de AI-transparantieverklaring?
Absoluut, en dit is zelfs een wettelijke verplichting. De AI-geletterdheidsplicht die sinds februari 2025 van kracht is, vereist dat medewerkers die met AI-systemen werken begrijpen hoe die systemen functioneren en welke invloed ze hebben op klantinteracties. In de praktijk betekent dit dat klantenservicemedewerkers moeten weten wat er in de transparantieverklaring staat, hoe ze klanten kunnen informeren over het AI-gebruik, en hoe ze een escalatie naar een menselijke medewerker correct afhandelen. Neem de verklaring daarom op in je onboarding- en trainingsmateriaal.
Kan ik één AI-transparantieverklaring gebruiken voor alle kanalen, of heb ik er per kanaal één nodig?
Je kunt werken met één overkoepelend document als basis, maar het is sterk aan te raden om per kanaal te specificeren welke AI-systemen actief zijn en hoe de klantmelding plaatsvindt. Een chatbot op je website, een voicebot in je IVR en een geautomatiseerd e-mailsysteem hebben elk een andere interactievorm en mogelijk ook andere gegevensverwerkingen. Door kanaalsspecifieke secties toe te voegen aan je hoofdverklaring, blijft het document zowel compliant als praktisch bruikbaar voor klanten die willen begrijpen wat er precies in hun specifieke contactmoment gebeurt.
Wat moet ik doen als een AI-systeem een foutieve beslissing heeft genomen die een klant heeft benadeeld?
Informeer de klant zo snel mogelijk, leg uit wat er is misgegaan en bied een correctie of compensatie aan. Documenteer het incident in je interne AI-register, inclusief de oorzaak, de gevolgen en de genomen herstelmaatregelen. Afhankelijk van de ernst van de situatie kan een melding bij de Autoriteit Persoonsgegevens of een andere toezichthouder verplicht zijn. Gebruik het incident ook als aanleiding om je AI-transparantieverklaring te herzien en te controleren of de beschreven rechten van klanten, zoals het aanvechten van een geautomatiseerde beslissing, in de praktijk ook daadwerkelijk uitvoerbaar zijn.
Hoe meet ik of mijn AI-transparantieverklaring effectief is en klanten daadwerkelijk bereikt?
Effectiviteit meet je op twee niveaus: bereik en begrip. Voor bereik kun je bijhouden hoeveel klanten de verklaring openen via directe links in je chatinterface of bevestigingsmails. Voor begrip is klantfeedback waardevoller: vraag in klanttevredenheidsonderzoeken of klanten wisten dat ze met een AI-systeem communiceerden en of ze wisten hoe ze een medewerker konden bereiken. Als een significant deel van je klanten aangeeft dit niet te weten, is dat een signaal dat je actieve melding of de zichtbaarheid van je documentatie moet verbeteren.
