European Commission sets AI transparency rules from 2 August

TL;DR: The European Commission said most of the EU AI Act's transparency rules will apply from 2 August, requiring AI companies to tell people when they are interacting with AI and to use machine-readable markings for AI-generated or AI-manipulated content. The Commission also said systems already on the market have until 2 December to comply with content marking and detection requirements.

The European Commission published guidance stating that most transparency obligations under the EU AI Act will take effect on 2 August. The guidance says AI companies must inform people when they are using AI and must apply machine-readable markings to help detect content that AI created or manipulated.

The European Commission said companies must also warn people when they are shown deepfakes and when they are shown AI-generated content about matters of public interest that does not involve human oversight. The same warning requirements apply to emotion recognition systems and biometric categorisation systems.

The European Commission said its guidance explains how to apply the transparency rules, including that direct interaction covers services such as chatbots. The Commission also said companies can use a code of practice to help demonstrate compliance with the AI Act's transparency requirements. It added that systems already available have until 2 December to meet the rules on content marking and detection, while enforcement powers will sit with the European Commission and relevant national authorities.

Related questions

• When do the EU AI Act transparency rules start to apply?
• What disclosures must AI companies provide under the European Commission guidance?
• Which AI systems must carry warnings under the new EU transparency rules?