AI Act: mandatory transparency from August 2 — act now

As of August 2, 2026, the transparency obligations under Article 50 of Regulation (EU) 2024/1689 apply across the EU. Users must be informed when interacting with a chatbot, and AI-generated or altered content must be clearly labeled.

Key facts

The European Commission released practical guidelines and a code of best practices to support compliance, notably for AI content labeling and user information. Competent authorities include national market surveillance authorities, the AI Office (for certain systems), and the EDPS for EU institutions.

For scope and requirements, see our overview of the AI Act and its core obligations.

Legal framework

• Article 50 — transparency: clear information when people interact with a direct-interaction AI system (chatbots, assistants), unless obvious.
• AI content labeling: human-readable notices and, where relevant, machine-readable metadata (e.g., C2PA) to enable automated detection.
• Emotion recognition/biometric categorization: information duties with limited exceptions (e.g., lawful criminal uses, editorial control with responsibility).
• Timeline: applicable since August 2, 2026; a limited grace period until December 2026 only for labeling of generative systems placed on the market before that date.
• GDPR interplay: GDPR transparency and information duties remain fully applicable.

What this means for Luxembourg companies

Any organization deploying a customer-facing or internal chatbot must display an explicit notice indicating AI interaction. Marketing, support, visuals/videos, audio and text generated/altered by AI must be visibly labeled and, where possible, enriched with robust metadata.

To structure your AI compliance program, our specialists can help you establish AI governance aligned with the AI Act and embed controls into editorial workflows.

Immediate actions

• Add “you are interacting with AI” notices across all touchpoints (web, mobile, voicebot, kiosks) and update scripts/UX.
• Implement dual labeling: visible icon/notice + machine-readable metadata (e.g., C2PA/Content Credentials) for all AI content.
• Define an exception and editorial control process, record human review and document editorial responsibility.
• Map deployments by entity/country (LU/BE/FR/DE) and contract labeling/metadata duties with providers.
• Update GDPR notices (Arts. 13/14) to align GDPR transparency with the AI Act notice.
• Keep evidence: screenshots, deployment timestamps, labeled content samples, internal audit reports and policies.

Need parallel GDPR alignment? Transparency and fairness remain essential; kick off with dedicated GDPR support in Luxembourg.

Bottom line

No general leniency applies: rules are in force now, with a limited grace period until December 2026 for certain legacy labeling. Prioritize chatbot notices, AI content labeling and robust evidence.

For a fast assessment and action plan, reach out via Luxgap’s contact page.

Article generated by Luxgap regulatory watch. For tailored guidance on this topic, contact us.