The EU AI Act and HR: Duties That Already Apply in 2026
Employers using AI in HR must already comply with three parts of the EU AI Act: AI literacy measures and the ban on workplace emotion recognition, both applying since February 2, 2025, and transparency duties applying since August 2, 2026. The heavier high-risk rules for hiring and worker management tools apply from December 2, 2027.
Key takeaways
- Article 4 requires providers and deployers to take measures to support AI literacy among staff who use AI systems.
- Article 5 bans AI that infers emotions of people in the workplace, except for medical or safety reasons.
- Article 50 transparency duties, such as making clear when people interact with AI, apply from August 2, 2026.
- Use the time before December 2, 2027 to inventory tools and prepare for high-risk deployer duties.
Which EU AI Act duties apply to HR right now?
Much of the discussion about the AI Act and HR focuses on high-risk systems, covered in detail in our guide to high-risk employment AI and deadlines. This article focuses on the duties that are already live, because they apply regardless of whether a tool is high-risk.
| Duty | Article | Applies from | What it means for HR |
|---|---|---|---|
| AI literacy | Article 4 | February 2, 2025 | Train and support staff who use AI tools |
| Prohibited practices | Article 5 | February 2, 2025 | No emotion inference on employees or candidates |
| Transparency | Article 50 | August 2, 2026 | Disclose AI interactions and emotion or biometric systems |
| High-risk deployer duties | Article 26 and Annex III | December 2, 2027 | Oversight, logs, worker notice for hiring and management AI |
The dates reflect Article 113 as amended by the 2026 Digital Omnibus, as published on the European Commission's AI Act Service Desk.
What does the AI literacy duty require?
As amended, Article 4 says providers and deployers must take measures to support the development of AI literacy of their staff and others who operate or use AI systems on their behalf. The amended text clarifies that it does not require employers to guarantee a specific level of literacy for each individual. In practice, regulators will expect you to show reasonable, documented effort.
- Identify which HR roles use AI tools: recruiters, HR business partners, payroll, HR operations and managers.
- Train each group on what the tools do, their limits, how to check outputs and how to escalate problems.
- Keep records of training content and completion.
- Refresh training when tools or uses change.
Our role-based AI literacy curriculum for HR teams gives a starting structure.
Which workplace AI practices are banned?
Article 5 prohibits using AI systems to infer the emotions of people in the workplace and in education, except where the system is intended for medical or safety reasons. For HR, review any feature described as sentiment analysis, engagement scoring from video or voice, or "honesty" and stress detection in interviews. If it infers emotions of identifiable employees or candidates in the EU, switch it off.
Fines for prohibited practices are the highest in the Act, so this check should not wait. Ask vendors directly whether any feature infers emotions, and get the answer in writing.
What do the transparency duties mean for HR chatbots?
Article 50 requires that AI systems interacting directly with people be designed so those people know they are dealing with AI, unless that is obvious to a reasonably well-informed person. The provider carries this design duty, but HR teams deploying an HR assistant should confirm the disclosure appears. Article 50 also requires deployers of emotion recognition or biometric categorisation systems to inform the people exposed to them, and to process personal data under the GDPR.
Fold these points into internal policy; our guide to writing an AI acceptable use policy for HR shows where they fit.
Starter checklist for HR
- Inventory every AI tool that touches EU employees or candidates, including features inside larger HR suites.
- Flag and disable any emotion inference features.
- Confirm chatbots and assistants disclose that users are interacting with AI.
- Document AI literacy training by role.
- Classify each tool for the December 2027 high-risk rules and ask vendors about their plans.
Frequently Asked Questions
Does the EU AI Act apply to US employers?
It can. The Act covers deployers established in the EU, such as a US company's EU subsidiary, and deployers outside the EU when the AI system's output is used in the EU. A US employer running global HR tools that make or inform decisions about EU candidates or employees should assume the Act may apply and seek legal advice.
Is an HR chatbot a high-risk AI system?
Usually not, if it only answers questions and does not make or materially influence decisions about individuals. It is still subject to Article 50 transparency, so users should know they are interacting with AI. If the chatbot screens candidates or recommends employment decisions, it may fall under the high-risk employment category instead.
What counts as AI literacy training?
There is no prescribed format. Training should help staff understand what each AI tool does, where it can be wrong, how to review outputs and when to escalate. Tailor content to roles, since recruiters and payroll teams use AI differently. Keep records of content, attendance and updates to show reasonable effort.
Next steps
This month, complete an AI tool inventory for EU-facing HR processes, disable any emotion inference features and confirm chatbot disclosures. Next quarter, document role-based AI literacy training and begin vendor conversations about the December 2027 high-risk requirements.
Last reviewed: October 2026.
Sources
- Article 4: AI literacy, European Commission AI Act Service Desk
- Article 50: Transparency obligations, European Commission AI Act Service Desk
- Article 113: Entry into force and application, European Commission AI Act Service Desk
Comments
Post a Comment