By Isabel Cortés. Head of Institutional Projects and Relations. Trademark Attorney PONS IP

The rapid growth of generative artificial intelligence has changed how digital content is created, shared and consumed. From AI-generated images and synthetic voices to highly realistic videos of events that never happened, these technologies are transforming online communication at an unprecedented limit. While generative AI offers significant benefits for creativity, innovation and efficiency, it also creates new risks, including misinformation, fraud and reputational harm. In response to these challenges, the European Union has introduced transparency obligations under the AI Act. On 10 June 2025, published the Code of Practice on Marking and Labelling of AI-Generated Content, which provides practical guidance on how AI-generated and AI-manipulated content should be identified and disclosed to users.

The Code of Practice supports the implementation of the AI Act’s transparency rules. Since 2 August 2026, organisations using certain AI systems must comply with requirements on the disclosure of deepfakes and some AI-generated content. The Code provides practical guidance on how these obligations can be met in a consistent way across the European Union.

Transparency Requirements Under the AI Act

The AI Act includes transparency rules to ensure that people know when they are viewing certain types of AI-generated content. Rather than prohibiting synthetic content, the Act focuses on informing users when AI has been used. The objective is to help people make informed decisions and maintain trust in digital information.

Article 50 requires disclosure of certain AI-generated or AI-manipulated content. This includes deepfakes, namely images, audio recordings or videos created or altered using AI that appear to show real people, places, objects or events and could be mistaken for authentic content. The rules may also apply to AI-generated text published on matters of public interest where there has been no human review or editorial oversight. In these situations, users must be informed that the content was generated or modified using AI.

As generative AI tools become more powerful and widely available, the ability to distinguish real content from synthetic content is becoming increasingly difficult. The EU sees transparency as a key safeguard to reduce misinformation and protect users.

The Code of Practice: A Practical Guide to Compliance

The Code of Practice provides practical guidance on how organisations can comply with the AI Act’s transparency obligations. It is aimed at providers, deployers, publishers and platforms that use generative AI systems.

Although the Code is voluntary, the underlying transparency obligations are not. The European Commission has made clear that using the recommended labels or icons does not automatically guarantee compliance. Organisations remain responsible for ensuring that their disclosures meet the requirements of Article 50 of the AI Act. However, the Code is therefore likely to become an important benchmark for companies seeking to demonstrate good compliance practices.

EU-Wide Labels for AI-Generated Content

One of the most visible features of the Code is the introduction of a common set of EU icons for labelling AI-generated content. These icons are intended to make it easier for users to recognise when content has been generated or manipulated by AI.

The Commission has developed three main categories:

• A basic AI icon, showing that AI was involved in creating or modifying content.
• A “Fully AI-Generated” icon for content created entirely by AI.
• A “Partially AI-Modified” icon for human-created content that has been altered using AI.

The creation of a common visual system across the EU may help reduce fragmentation between platforms and improve recognition by users. User testing carried out by the Commission also showed that labels are more effective when the icon is accompanied by clear explanatory text.

Display Requirements and Exceptions

The Code recommends that disclosures should be clearly visible when users first encounter the content. Where possible, labels should be embedded directly into the content and remain visible when it is shared or downloaded. It also encourages accessibility measures such as clear language, readable sizing and compatibility with assistive technologies.

The AI Act also provides important exceptions. Disclosure requirements are more flexible for content that is clearly artistic, fictional, creative or satirical. They do not apply when the use of deepfakes is authorised by law for criminal investigations. In addition, AI-generated text may be exempt where it has undergone human review, and a person or organisation assumes editorial responsibility for the publication.

What Businesses Should Do Next and the power of Trade Marks

The publication of the Code of Practice marks an important step in the EU’s approach to AI transparency. Technology companies, media organisations, content creators and online platforms should assess whether their AI-generated content falls within the scope of the transparency obligations and whether appropriate labelling processes are in place.

Companies operating across the EU should not view the Code simply as a communications tool. Although voluntary, it is likely to become an important tool for assessing compliance with Article 50 of the AI Act and may be used by regulators as evidence of good practice. As AI-generated content becomes increasingly difficult to distinguish from reality, transparency is rapidly becoming both a compliance requirement and an essential element of responsible AI governance.