EU AI Act Article 50 Is Live: How Brands Should Label AI Video

Over the past year, AI video has gone from a fun experiment to an everyday production tool for brands. But since 2 August 2026, there is a new rule in the game: the transparency obligations in Article 50 of the EU Artificial Intelligence Act now apply. If your videos reach the EU market, AI video labeling is no longer just a matter of brand self-discipline; it is a legal responsibility. This article explains the key points in terms brand owners can use, along with a workflow you can adopt right away.

This article is general information, not legal advice. Please consult a lawyer familiar with EU regulation for your specific situation.

From 2 August: what Article 50 of the EU AI Act covers

According to the European Commission's guidance, Article 50 is about making sure people know when they are dealing with AI. Its main obligations include:

• AI systems that interact with people, such as chatbots, must let users know they are talking to AI, unless that is already obvious from the context.
• Providers of generative AI must ensure that generated images, video, audio and text carry machine-readable marks that are "effective, reliable, robust and interoperable", so the content can be detected as AI-generated or manipulated.
• Deployers who use deepfake content must disclose, clearly and distinguishably and at the latest at the viewer's first exposure, that the content was artificially generated or manipulated.

On timing, Article 50 has applied since 2 August 2026; for systems already on the market, the machine-readable marking obligation has a grace period only until 2 December 2026. Fines can reach 15 million euros or 3% of total worldwide annual turnover, with proportionality taken into account for SMEs.

Tools mark the content, brands disclose it

Many brands assume this is the AI tool vendor's problem. In fact, Article 50 splits responsibility into two layers:

• Providers: companies that build AI video tools are responsible for making every output carry a machine-readable mark, using techniques such as watermarks or content credentials.
• Deployers: those who use these tools and publish the content, meaning brands and agencies. Whenever a video contains deepfake content that looks or sounds like real people, places or events, they must actively disclose it.

To give the industry a consistent approach, the EU finalized the Code of Practice on Transparency of AI-generated Content on 10 June 2026. By the end of July, around 190 companies and organizations had signed it, and the Commission and the AI Board confirmed it as an adequate tool for demonstrating compliance. The EU has also designed a set of icons that deployers can use to label AI-generated content.

In other words, choosing the right tool only gets you halfway. For every video your brand publishes, the duty to disclose sits with the brand.

The creative-work exception does not mean no label

Brand ads are often fictional stories and exaggerated creative ideas. Does that make them exempt? The answer: the rules relax, but they do not disappear.

Under Article 50, when deepfake content is part of an "evidently artistic, creative, satirical, fictional or analogous work", disclosure can take an appropriate form that "does not hamper the display or enjoyment of the work", such as an end card or description field rather than large text across the middle of the frame. But the disclosure itself must still be there.

Also note that AI-generated "spokespeople", voices that imitate real people, and photorealistic recreations of real scenes can easily fall within the scope of deepfakes. Flag these assets at the planning stage rather than scrambling to fix them before launch.

An AI video labeling workflow for brands

We recommend building labeling into production rather than adding a sticker at the end:

1. Keep an asset log. For each video, record which AI tools were used and which shots or sounds were generated.
2. Choose tools that can mark content. Prefer platforms that automatically add machine-readable marks or content credentials, and keep the original output files.
3. Grade the risk. Distinguish "AI-assisted editing and color grading" from "generating realistic people, voices and scenes"; the latter always goes through disclosure.
4. Standardize the disclosure format. Write label placement, wording and icons into your brand guidelines and apply them consistently across markets and platforms.
5. Human review before publishing. Have someone who knows both the brand and the rules make the final check, confirming disclosure is in place and nothing is misleading.

The division of labor between people and machines is especially clear here:

• AI handles generating assets, automatically adding machine-readable marks and keeping generation records.
• People handle deciding what counts as a deepfake, choosing how to disclose, taking final responsibility for publishing, and making sure labels do not spoil the work's aesthetics.

If you are redesigning your AI video workflow, see our earlier article, What Sora's Shutdown Teaches Brands, on avoiding betting your workflow on a single model.

Conclusion

AI video labeling may look like a compliance burden, but seen another way, it is a chance for a brand to show sincerity. As AI content multiplies, brands willing to say clearly "we used AI here" are more likely to earn trust. AI makes it perfect, people make it moving, and honesty is one of the most moving qualities a brand can have. If you want an AI video workflow that balances creativity, efficiency and compliance, explore TitApeX services or talk to us.

References

• Transparency obligations under Article 50 of the AI Act (European Commission)
• Code of Practice on Transparency of AI-generated Content (European Commission)
• Guidelines on transparency obligations for providers and deployers of certain AI systems (European Commission)
• Deepfakes, Chatbots, AI-Generated Text: European Commission Details Transparency Obligations Under the AI Act (Greenberg Traurig, June 2026)