The classic trap
Recital 92 reminds employers of something many Luxembourg companies discover too late: deploying a high-risk AI system (candidate scoring, productivity monitoring, task allocation, performance review) without first informing the staff delegation and the workers concerned is a serious breach. The CNPD has already intervened on such systems under Article 88 GDPR, and the staff delegation can refer the matter to the ITM. Recital 92 closes the gap: even when Directive 2002/14/EC or the Luxembourg Labour Code does not apply (small structures, types of decisions not covered), the AI Act imposes a stand-alone information duty.
What this recital concretely changes for your HR AI projects
Recital 92 informs the reading of Article 26(7) AI Act and forces a redesign of AI project governance from the scoping phase. Practical points for Luxembourg:
- Information must be prior to deployment, not concurrent: an email sent on go-live day is not enough.
- The staff delegation (Labour Code L.414-3 and L.414-9) must be consulted whenever AI affects work organisation, activity monitoring or employment conditions.
- Information targets workers AND their representatives: dual channel is mandatory.
- Minimum content: purpose of the AI system, categories of decisions, general logic, workers' rights (Article 86 AI Act on explanations).
- The AI Act stacks with Article 88 GDPR and Article L.261-1 of the Luxembourg Labour Code on workplace surveillance, which requires prior CNPD authorisation for certain processing.
- Defective information weakens any HR decision taken through AI (dismissal, denied promotion, rejected hire) before the Labour Court.
How Luxgap automates this risk
Our Luxgap Workforce AI Disclosure Agent turns the Article 26(7) information obligation into an automated and enforceable workflow. The LLM agent detects, across your connected HR tools (Workday, Sopra Steria HR Suite, SAP SuccessFactors, BambooHR, Lucca, Cegid Talentsoft, Microsoft Viva Insights), AI systems that are active or about to be deployed and that influence decisions on workers, then automatically produces the information package tailored to the Luxembourg context.
- Continuously scans your HR suites and productivity platforms to identify active AI features (CV scoring, training recommendations, turnover detection, performance monitoring).
- Qualifies each feature against Annex III of the AI Act and determines whether it is high-risk or not, with written justification.
- Generates the bilingual FR/EN information pack for the staff delegation: purpose, logic, data used, Article 86 rights, human oversight measures.
- Produces a pre-filled minutes template for the delegation meeting, aligned with Articles L.414-3 and L.414-9 of the Labour Code.
- Alerts the DPO and CISO in real time whenever a new AI module is activated in an HR tool without documented prior information.
- Archives a time-stamped and cryptographically signed PDF, enforceable before the ITM, the CNPD and the Labour Court in case of dispute.
Available as a complement to a Luxgap DPO mandate or as a dedicated SaaS module depending on your scope. Request a demonstration on your real HR tools, with a free 48-hour blank audit to map the AI systems already deployed at your company and assess legal exposure before any engagement.
