15/03/2026
The EU is speeding up the implementation of the AI Act with new rules on the safety and labelling of AI-generated content
By Iván Martínez, CEO of Intedya
The development of the European AI Act continues to move towards its practical implementation phase. One of the key issues shaping the regulatory agenda is the safety of artificial intelligence systems and enforcement mechanisms (application and supervision of the law). In this context, two recent initiatives have taken on particular significance: the publication of the International AI Safety Report 2026 and the new draft Code of Good Practice for the labelling and tagging of AI-generated content.
AI safety and international cooperation
The International AI Safety Report 2026 was presented during the New Delhi AI Impact Summit, an international conference focusing on the global impact of artificial intelligence. The report analyses the main risks associated with the development and deployment of advanced AI systems and proposes frameworks for international cooperation to improve their governance. Carina Prunkl, the report’s lead author, emphasised that the study’s findings are particularly relevant to the implementation of the European Union’s AI Act, as they provide technical evidence and recommendations that can support the development of standards for the safety, auditing and oversight of high-risk systems. The report highlights the need to strengthen mechanisms for independent assessment, continuous monitoring and algorithmic transparency.
New draft Code of Good Practice on AI-generated content
At the same time, the European Commission has published the second draft of the Code of Good Practice on the labelling and tagging of AI-generated content, a voluntary initiative designed to facilitate compliance with Article 50 of the AI Act, which sets out transparency obligations for certain AI systems.
The document incorporates feedback received from hundreds of stakeholders, including technology companies, universities, civil society organisations, Member States and representatives of the European Parliament. This participatory process took place throughout January 2026 through consultations, meetings and technical workshops.
The new draft aims to reduce the compliance burden on organisations whilst promoting open labelling standards and proposing the creation of a common European icon to clearly identify content generated by artificial intelligence.
The code is structured into two main sections:
- Section 1 – Providers of generative AI systems: this addresses mechanisms for flagging and facilitating the automatic detection of AI-generated content, incorporating greater technical flexibility to accommodate different types of models and platforms.
- Section 2 – Deployers or users of AI systems: this focuses on labelling obligations for content such as deepfakes or AI-generated text when it relates to matters of public interest, with a more practical approach geared towards actual implementation.
Next regulatory steps
The European Commission has opened a public consultation on this second draft, which will remain open until 30 March 2026. Following an analysis of the contributions received, the final version of the code is expected to be published in early June 2026. The transparency rules relating to AI-generated content set out in the AI Act will apply from 2 August 2026, meaning that technology companies, digital platforms and organisations using generative AI will need to prepare their content identification and labelling mechanisms in advance.
These developments reflect the AI Act’s transition from the legislative phase towards its operational implementation, where aspects such as security, transparency and the effective oversight of AI systems become key elements. For organisations that develop or use artificial intelligence, the current period presents an opportunity to adapt their systems, compliance processes and governance models to the new European regulatory requirements.
