Guidelines on Transparency Obligations for Providers and Deployers of Certain AI Systems

Applies To

Overview

These guidelines were issued by the European Commission on 20 July 2026 to support practical compliance with Articles governing transparency under the EU AI Act, ahead of the 2 August 2026 application date for those provisions. The guidelines address four principal areas: disclosure obligations when users interact with AI systems, labeling of AI-generated content, record-keeping and logging duties, and the internal governance workflows required to authorize AI-assisted outputs. They apply to any provider placing an AI system on the EU market and any deployer using such a system within the EU, regardless of where the organization is headquartered. While the guidelines do not create new legal obligations beyond those in the AI Act itself, they carry significant interpretive authority and regulators are expected to reference them in enforcement proceedings. Organizations that fail to implement the practices described risk being found in breach of transparency provisions that carry fines of up to 15 million euros or 3 percent of global annual turnover under the AI Act enforcement regime.

Key Requirements

• •Providers and deployers must inform users when they are interacting with an AI system, except where this is obvious from context, effective from 2 August 2026.
• •AI-generated content, particularly synthetic audio, video, images, and text intended for public dissemination, must be labeled in a machine-readable format and with a visible disclosure.
• •Deployers of emotion-recognition or biometric-categorization systems must notify individuals subject to those systems before or at the point of use.
• •Providers must maintain technical documentation and logs sufficient to demonstrate compliance with transparency obligations throughout the system lifecycle.
• •Internal approval workflows must be established and documented to govern the release of AI-generated or AI-assisted outputs, with records retained for a minimum period aligned with AI Act requirements.
• •Non-compliance with transparency obligations may trigger fines of up to 15 million euros or 3 percent of worldwide annual turnover, whichever is higher.

What Your Organization Must Do

• →Audit all AI systems currently in use or under development to identify which generate or assist in producing user-facing outputs, and map each to the relevant transparency obligation.
• →Implement disclosure mechanisms, including on-screen notices and machine-readable labels, for any AI-generated content before the 2 August 2026 deadline.
• →Update product documentation, terms of service, and user interface flows to include legally compliant AI interaction notices as specified in the guidelines.
• →Establish or revise internal approval workflows for AI-assisted content release, ensuring sign-off procedures are documented and auditable.
• →Configure logging and record-retention systems to capture the data required to demonstrate transparency compliance across all covered AI systems.
• →Update vendor and partner contracts to confirm that third-party AI systems supplied to your organization meet the labeling and disclosure standards set out in the guidelines.

Playbook Guidance

Step-by-step implementation guidance for compliance teams.

Frequently Asked Questions

Do the EC transparency guidelines apply to companies headquartered outside the EU if they deploy AI systems within the EU?

Yes. The guidelines apply to any provider placing an AI system on the EU market and any deployer using such a system within the EU, regardless of where the organization is headquartered. Non-EU companies operating in this space should treat these guidelines as directly applicable to their EU-facing products and services.

What is the compliance deadline for the EU AI Act transparency obligations covered by these guidelines?

The transparency provisions of the EU AI Act apply from 2 August 2026. The European Commission published these guidelines on 20 July 2026 specifically to support organizations in achieving compliance before that date, leaving a narrow implementation window for any remaining gaps.

What fines apply if an organization fails to meet the AI Act transparency obligations described in these guidelines?

Non-compliance with transparency obligations can trigger fines of up to 15 million euros or 3 percent of worldwide annual turnover, whichever is higher. Regulators are expected to reference these guidelines in enforcement proceedings, making adherence to their practices important for limiting liability.

Do the EC transparency guidelines create new legal obligations beyond what the EU AI Act already requires?

The guidelines do not create new legal obligations, but they carry significant interpretive authority. Because regulators are expected to use them as a benchmark in enforcement proceedings, departing from their recommended practices substantially increases the risk of being found in breach of the underlying AI Act provisions.

What labeling format is required for AI-generated content under these guidelines?

AI-generated content intended for public dissemination, including synthetic audio, video, images, and text, must carry both a machine-readable label and a visible disclosure to users. Organizations should implement both formats before the 2 August 2026 application date to satisfy the dual-format requirement.

What internal governance requirements do the guidelines impose on deployers of emotion-recognition or biometric-categorization AI systems?

Deployers of these systems must notify individuals subject to them before or at the point of use. Beyond that individual-facing obligation, organizations must also establish documented internal approval workflows for AI-assisted outputs and maintain auditable records, with retention periods aligned to AI Act requirements.