EU Transparency Requirements for AI Content: What Applies from August 2026

The European Commission has published the first draft of a Code of Practice for labeling AI-generated content. From August 2, 2026, providers and operators of AI systems must comply with strict transparency rules – from machine-readable watermarks to the labeling of deepfakes. For companies, the clock is now ticking.

EU-Transparenzpflicht für KI-Inhalte: Was ab August 2026 gilt

English edition — originally published in German as EU-Transparenzpflicht für KI-Inhalte: Was ab August 2026 gilt.

# EU Transparency Obligation for AI Content: What Applies from August 2026

The European Commission has published the first draft of a code of practice for labeling AI-generated content. From August 2, 2026, providers and operators of AI systems must comply with strict transparency rules – from machine-readable watermarks to the labeling of deepfakes. For companies, the clock is now ticking.

New Rules for AI-Generated Content

With the publication of the first draft of the Code of Practice on Transparency for AI-Generated Content, the EU Commission sets a decisive milestone in the implementation of the AI Act. The regulations aim to make AI-generated content – from synthetic images to deepfakes to automatically generated texts – transparent and comprehensible for users.

The transparency obligation affects two central actors: AI Providers (providers of AI systems) and AI Deployers (operators who use AI systems). While providers must implement technical labeling, deployers are responsible for making this information visible to end-users – especially in sensitive applications.

According to Kirkland & Ellis, the code includes detailed specifications for watermarks, metadata, and disclosure mechanisms. The regulations will come into force simultaneously with the high-risk provisions of the AI Act: on August 2, 2026. Companies therefore have about six months left for implementation.

Technical Requirements: Watermarks and Metadata

The Code of Practice obliges AI Providers to provide synthetic content with machine-readable labels. These must be robust against manipulation and at the same time consider standards such as C2PA (Coalition for Content Provenance and Authenticity).

Specifically, this means:

Particularly relevant: The technical solutions must be interoperable to function across different platforms and systems. This poses a technological challenge, especially for smaller providers.

Disclosure Obligations for Deployers: When Must Content Be Labeled?

While providers are responsible for the technical infrastructure, AI Deployers must ensure that users are informed about AI-generated content. The code differentiates according to risk categories:

Mandatory labeling applies to:

The EU Commission emphasizes in its press release that the regulations are deliberately risk-based. Not every AI-generated newsletter or chatbot output needs to be explicitly labeled – the context and potential for misuse are decisive.

Companies should therefore already establish processes to identify relevant AI applications and evaluate labeling obligations.

Interaction with the AI Act: Governance Across the Entire Lifecycle

The transparency obligation is not an isolated set of rules, but an integral part of the broader AI Act governance requirements. As current analyses show, the AI Act demands lifecycle-wide governance – from design to deployment to post-market monitoring.

Specifically, this means:

In parallel, the debate about the Digital Omnibus is ongoing – a package to simplify the AI Act, which the EU Commission and Parliament are currently controversially discussing. While industry calls for simplifications, civil rights organizations warn against weakening protection standards. Companies should keep an eye on these developments, as they could impact compliance requirements.

Also interesting is a look at Switzerland: The termination of the Palantir contract due to data sovereignty risks shows that regulatory and strategic considerations regarding AI governance are also increasing beyond the EU. Companies in the DACH region are operating in an increasingly complex regulatory environment.

Practical Relevance: What Companies Should Do Now

With the deadline of August 2, 2026, companies have just under six months for implementation. Those who act now will not only avoid fines (up to 35 million euros or 7% of global annual turnover) but also position themselves as responsible AI users.

Concrete steps for action:

  1. Inventory: Recording all AI systems that generate content (internally and externally used tools)
  2. Risk classification: Assessing which systems fall under the disclosure obligation
  3. Technical implementation: Integration of watermarks, metadata, or other labeling solutions
  4. Process design: Establishing workflows to consistently implement labeling
  5. Training: Raising awareness among teams that use or operate AI tools
  6. Monitoring: Setting up control mechanisms for continuous compliance monitoring

Particularly critical: The regulations affect not only large tech corporations but all companies that use AI-generated content in communication, marketing, or products. A medium-sized company that generates personalized product descriptions via GPT-4 falls under the regulation just as much as a SaaS provider with synthetic avatars in customer support.

Conclusion: Transparency as a Strategic Competitive Factor

The EU transparency obligation for AI-generated content is more than a regulatory hurdle – it is a signal for responsible handling of generative AI. Companies that understand transparency not only as a compliance requirement but as a trust factor gain a competitive advantage.

The coming months will be crucial: In parallel with the implementation of the transparency rules, the deadline for high-risk systems is running, and the political debate about simplifications remains dynamic. Those who build a robust AI governance structure early on will be prepared for all scenarios.

Deep Impact AG has been supporting companies in the DACH region since 2017 in developing AI strategies and governance frameworks that meet regulatory requirements while maintaining innovative strength. From risk classification to technical implementations and training your teams – we accompany you on the path to AI Act compliance. Contact us for a non-binding initial consultation at deep-impact.ch.