Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems

Specific requirement to comply with immediately

The European Commission's guidelines state that from 2 August 2026 providers and deployers must meet the AI Act's transparency obligations so people can recognise AI interactions and AI-generated or altered content (according to the European Commission: digital-strategy.ec.europa.eu). The guidelines distinguish obligations for interactive AI systems and for marking AI-generated content.

Which transparency obligations the AI Act imposes on providers and deployers

The guidelines clarify two main obligation categories: informing users when they directly interact with AI, and labelling content produced by AI. The Commission emphasises obligations apply to system designers (providers) and entities deploying systems (deployers). Clear notices or labels and mechanisms to prevent deception are required (per the European Commission).

What technical steps a company should take before requirements start

1. Audit interactive and content-generating functions to identify points where users may not recognise AI.
2. Implement labelling mechanisms for messages and content (for example metadata, visible banners, explicit labels) and event logging for inspection.
3. Provide the ability to disable automated features where the risk of misinformation is high.

Each step follows the Commission's recommendations on technical measures and activity recording.

What organisational and procedural steps companies must implement

1. Update information security policies and change-management procedures to include AI labelling requirements.
2. Train product and customer-facing teams to communicate AI involvement to users.
3. Appoint a responsible person for AI Act compliance and establish audit and incident-escalation processes.

The Commission's guidelines stress internal accountability and evidencing compliance for potential checks.

Where the main pitfalls and risks lie for businesses

Common pitfalls include lacking a central inventory of AI uses, unexpected content generation via chained models, and difficulties labelling when integrating external APIs. Failure to record activities will make demonstrating compliance in audits difficult (according to the European Commission). Industry reporting also notes scaling challenges for smaller companies.

Summary

Companies operating in the EU should, by 2.08.2026, consolidate an AI inventory, deploy technical labelling and logging, and update organisational procedures. These measures follow directly from the European Commission's guidelines and form a practical minimum plan before the obligations take effect.

Lub System helps B2B companies implement AI, automation and IT solutions end-to-end - from strategy to deployment. See our services or get in touch to discuss your case.