The European Commission on Wednesday published a practical code of practice for labeling and marking content created or modified by artificial intelligence (AI). The document is intended to help providers and organisations using AI systems prepare for the EU transparency requirements that take effect on 2 August 2026.
What the code covers
The code offers detailed practical and technical guidance on how to clearly indicate when content has been produced or altered with the assistance of AI. Key points include:
- clear labeling of deepfake-style videos and audio recordings;
- visible and easily interpretable markings for AI-generated or manipulated text;
- informing users when they interact with interactive AI systems, such as chatbot services.
The guidance outlines technical and organisational measures generative AI providers can adopt to ensure labels are consistent and understandable for end users.
Focus on generative systems and public-interest content
The document specifically addresses obligations for operators of generative AI systems. It provides instruction on labeling deepfakes and on marking AI-generated or AI-modified texts that are published in relation to matters of public interest, particularly when such materials have not undergone human review or editorial oversight.
Objectives and Commission encouragement
According to the Commission's statement, the transparency requirements aim to make it easier for people to recognise when they encounter AI-produced or AI-altered content, thereby reducing the risk of deception and manipulation. The Commission also encourages service providers and operators to sign up to the code to promote responsible, transparent and safe development and use of AI across the European Union.
Entry into force
The code is designed to help stakeholders comply with the forthcoming EU transparency rules, which enter into force on 2 August 2026. The Commission's practical guidance is intended to assist actors in adapting to the new requirements in time.
Summary
The European Commission's code provides concrete tools and principles for clearly marking AI-created content, with particular emphasis on deepfakes and generative AI outputs. The initiative aims to protect users and foster more transparent, responsible use of AI within the EU.
