Good news: Standard Zendesk customer support bots are limited risk — you need one clear disclosure message, not a full compliance programme. This guide covers what to add, when it escalates to high-risk, and the August 2026 deadline.
Deadline: 2 August 2026. Article 50 has NOT been deferred. Chatbot disclosure is required from this date regardless of other EU AI Act timeline changes.
Zendesk AI EU AI Act Compliance: What to Add Before August 2026
Zendesk AI bots serve millions of EU customers daily. Under Article 50 of the EU AI Act, any AI system that interacts directly with EU users must disclose its AI nature. Here is exactly what to configure in Zendesk — and when you face bigger obligations.
Zendesk AI features and EU AI Act classification
Feature
Classification
Obligation
Answer Bot / AI agent — automated responses
LIMITED RISK
Art.50(1) — disclose AI nature at conversation start
Zendesk messaging bot — chat widget
LIMITED RISK
Art.50(1) — disclose AI nature at conversation start
AI-generated ticket summaries (agent-facing)
NOT HIGH RISK
Internal tool, users not directly interacting with AI
Zendesk AI for credit / benefits / employment decisions
HIGH RISK
Annex III — full deployer obligations apply
What to add to Zendesk — takes under 1 hour
Add an AI disclosure as the bot's first message before any user input
Ensure the disclosure is in the same language as the conversation — configure per-language variants if you serve multiple EU markets
Update your privacy policy to reference AI chatbot use in customer interactions
Retain conversation logs in line with your data retention policy (Zendesk logs by default — verify your plan includes retention)
Copy a disclosure into your Zendesk bot's opening step:
Hi! I'm an AI assistant. I can answer most questions right away. Type "agent" or "human" to speak with a person.
Hello — you're chatting with an automated AI. A support agent is available if needed, just ask.
Need disclosure in French, German, Spanish, Dutch, Polish, Italian, Portuguese?
The Chatbot Compliance Pack includes 7 EU language variants, a T&C AI clause, privacy policy section, and machine-readable marking guide for AI-generated content.
Zendesk is a support platform that can be configured for many purposes. In most deployments — answering product questions, handling returns, routing tickets — it is limited risk. It becomes high-risk when the bot's outputs influence Annex III decisions:
Financial services: A Zendesk bot that determines whether a customer is eligible for a credit limit increase or insurance claim payout
Benefits / public services: A bot that screens entitlement questions and determines access to benefits
Employment: A Zendesk bot used as a first-stage HR screening tool for job applicants
If your Zendesk configuration falls into any of these categories, contact us — the compliance requirements are substantially heavier than a simple disclosure message.
Frequently asked questions
We are a US company with EU customers using our Zendesk chat. Does the EU AI Act apply?
Yes. Article 2(1)(c) applies the EU AI Act to providers and deployers established in third countries when their AI system's output is used in the EU. Your Zendesk bot chatting with EU users is exactly this scenario. The disclosure obligation applies regardless of where your company is headquartered.
Our bot has a human-sounding name (e.g. "Sophie"). Is that allowed?
A human-sounding name alone does not violate Article 50 — the obligation is to disclose the AI nature, not to use a robotic name. However, if "Sophie" is presented in a way designed to make users believe they are talking to a human employee (human photo, "our specialist Sophie will help you"), that deceptive design is prohibited. Add an explicit AI disclosure in the opening message regardless of the name.