Skip to main content

EU AI Act Article 13: Transparency Requirements

What EU AI Act Article 13 requires providers to disclose in instructions for use — performance metrics, limitations, and human oversight measures.

Transparency and provision of information to deployers

Article 13 requires high-risk AI systems to be sufficiently transparent so deployers can interpret outputs and exercise proper human oversight. Providers must supply comprehensive instructions for use.

Required instructions for use

Provider identity and contact details

Include the provider's name, registered trade name or trademark, postal address, and contact email. Deployers must know who built the system.

Characteristics, capabilities, and performance limitations

Document accuracy metrics, robustness thresholds, cybersecurity measures, and performance across demographic groups. Include all known limitations.

Foreseeable circumstances that may lead to risks

Describe situations where the system may underperform, produce biased outputs, or fail. Cover edge cases and out-of-distribution scenarios.

Changes to the system predetermined by the provider

Disclose any automatic updates, retraining schedules, or configuration changes the provider may apply after deployment.

Human oversight measures and maintenance

Explain technical measures that enable human oversight per Article 14, expected system lifetime, and required maintenance or care procedures.

Implementation steps

Generate instructions from model cards and datasheets

Automate instructions-for-use generation by extracting metadata from model cards, datasheets, and evaluation reports. Keep them version-controlled alongside code.

Version limitation disclosures and update logs

Track every known limitation, failure mode, and edge case in a structured changelog. Update instructions whenever the system changes.

Document demographic performance breakdowns

Measure and publish accuracy, precision, recall, and fairness metrics segmented by protected attributes and use-case scenarios.

Provide technical measures for output interpretation

Include confidence scores, explainability features, and guidance on when human review is required. Link to Article 14 oversight procedures.

Provider-to-deployer transparency chain

Article 13 creates a legally binding transparency chain: providers document system behavior, deployers receive instructions, and deployers use them to exercise human oversight under Article 14. If instructions are incomplete, deployers cannot comply with their own obligations.

Frequently asked questions

Related articles


How Scanara Helps

Scanara automates EU AI Act compliance from code to dossier. Connect your GitHub repos and get compliance reports in minutes.