Since 2 August 2026, the transparency obligations under the EU AI Act have applied. They affect every company whose employees use AI chatbots for work purposes. This is because the company itselfânot the individual employeeâis considered the deployer under the AI Act.
đ The key requirements of Article 50 of the AI Act in the HR context:
âĄď¸ Published AI-generated texts on matters of public interest must be labelled using a symbol and/or a notice.
⥠However, this requirement does not apply if the AI-generated texts have undergone substantive human reviewânot merely a spelling checkâand have been revised where necessary, provided that the company assumes editorial responsibility for them.
âĄď¸ Deepfakes must always be disclosed, including in job advertisements and employer-branding content. A notice in the legal disclosure or imprint is not sufficient.
âĄď¸ The use of emotion-recognition and biometric-categorisation systemsâfor example, in HR toolsâmust be disclosed to the individuals concerned (Article 50(3) of the AI Act).
â To-dos for employers:
âĄď¸ Create an inventory of AI systems and identify relevant workflows within HR management.
âĄď¸ For each AI system in use, determine whether the company qualifies as a provider or a deployer, as the respective obligations differ considerably.
âĄď¸ Train employees in accordance with their roles and responsibilities. The AI-literacy obligation under Article 4 of the AI Act has already applied since February 2025.
âĄď¸ Define internal responsibilities and ensure appropriate documentation.
â ď¸ Breaches of the transparency obligations may result in fines of up to âŹ15 million or 3% of worldwide annual turnover, as well as cease-and-desist claims under competition law.
P.S. The first draft of this post was written by Claude (Anthropic). Since I substantially revised it and assume editorial responsibility for it, THIS notice would not be required under Article 50 of the AI Act. đ