European Commission details AI Act transparency rules from 2 August

The European Commission said most AI Act transparency rules will apply from 2 August, with existing systems given until 2 December to comply with content marking and detection requirements.

TL;DR

• The European Commission published guidance on AI transparency obligations ahead of the AI Act taking effect on 2 August.
• Companies must tell people when they are interacting with AI and use machine-readable marks for content made or manipulated by AI.
• Deepfakes, AI-generated public-interest content without human oversight, and emotion recognition or biometric categorisation systems must carry warnings.
• Existing systems have until 2 December to meet marking and detection requirements.

The European Commission published guidelines explaining that companies must inform people when they are dealing with artificial intelligence and must apply machine-readable marks to identify content created or manipulated by AI. The guidance is part of the next implementation stage of the EU AI Act.

The European Commission said the transparency requirements also apply when people are shown deepfakes, AI-generated content involving matters of public interest where there is no human oversight, and systems used for emotion recognition or biometric categorisation. The guidance also states that direct interaction includes services such as chatbots.

The European Commission said most of the AI Act will be implemented on 2 August, including enforcement powers for the Commission and relevant national authorities. The Commission also pointed companies to a code of practice it said provides legal certainty on meeting the transparency requirements.

Related questions

• When do the EU AI Act transparency rules start to apply?
• What AI content must be marked under the European Commission guidance?
• Which existing AI systems have until 2 December to comply?