Sector
AI literacy training for customer service and retail.
Chatbots, ticket triage and personalisation engines talk to your customers every day. Article 4 covers the staff who run them, and your measures should help them understand what those systems do. Article 50 separately requires that customers know when they are talking to a machine.
Where AI is already in the work
- Customer-facing chatbots and voice assistants
- Ticket triage, routing and suggested replies in helpdesk software
- Personalisation, recommendation and dynamic pricing engines
- Call transcription, summarisation and quality scoring
- Fraud and returns-abuse flags at checkout
What the AI Act asks of you
Article 4 applies to every organisation whose staff use an AI system at work, and it has applied since 2 February 2025. Supervision and enforcement by national authorities started on 2 August 2026. The duty is to take measures that support the development of AI literacy among your staff and the people using AI on your behalf, calibrated to the tools in use, the people using them and the context. The Commission's Q&A is explicit that no certificate is required and that organisations can keep an internal record of training and guidance.
Article 50 requires that people interacting with an AI system are informed that they are doing so, unless it is obvious from the circumstances. That duty applied from 2 August 2026. Where a system infers emotions from customers, Annex III point 1 lists emotion recognition as high-risk and Article 50 requires the people exposed to it to be told.
Call quality scoring and agent monitoring features can also fall under Annex III point 4, AI used to monitor and evaluate the performance and behaviour of workers, with deployer obligations from 2 December 2027. A system already in service before that date is brought in once its design changes significantly. The tool inventory separates the customer-facing features from the staff-facing ones, and both are inside the Article 4 measure now.
Which workshop fits
Retail and service operations usually run the half day for frontline and supervisory staff, with the full day for the people who configure the bot, the routing rules or the monitoring features.
This is general information about the EU AI Act, not legal advice. Enforcement is proportionate and case by case, and your sectoral regulator may issue guidance that applies to you.
Questions from this sector.
Does our chatbot have to say it is a bot?
Yes, unless it is obvious from the circumstances. Article 50 requires that people interacting with an AI system are informed that they are, and that duty applied from 2 August 2026.
Is agent quality scoring high-risk AI?
If it uses AI to monitor and evaluate the performance and behaviour of workers, Annex III point 4 lists that as high-risk, with deployer obligations from 2 December 2027, or once the design changes significantly for a system already in service before then. Article 4 applies to it now.
Our staff only use suggested replies. Is that really AI use?
Yes. A suggested reply is generated by an AI system and sent under your authority. The Article 4 measure for that team is short and practical: what the feature does, where it goes wrong, and what the agent's own check adds.
Related insights.
Scope
Does the EU AI Act apply to us if staff only use ChatGPT?
Yes. The European Commission's Q&A says a company whose employees use ChatGPT for tasks like writing or translating must comply with Article 4, and should inform staff about specific risks such as hallucination.
Sectors
We did not choose the AI in our HR software. Are we still a deployer?
Yes. You are a deployer of an AI system when you use it under your authority, whether or not you chose the feature. The Article 4 duty follows the use, not the purchase decision.