The European Commission has published a Code of Practice on marking and labeling AI-generated content, giving organizations that deploy generative AI practical guidance for meeting transparency obligations connected to the EU AI Act. The code addresses the labeling and detection of AI-generated material, including deepfakes, and provides specific guidance for deployers of generative AI systems.

For businesses using AI to produce public-facing content or operate customer chatbots, the immediate issue is not simply whether a tool uses AI. It is whether people can understand when content has been generated or materially manipulated by AI, and whether that information is presented consistently. The Commission's official announcement on the Code of Practice positions the framework as a way to facilitate compliance and consistency across services and platforms.

The supplied Commission materials place the code's applicability around the third quarter of 2026. It is therefore a useful operational reference point for companies reviewing how they use generative AI in marketing, support, publishing, and other customer-facing workflows.

What the EU AI labeling code means in practice

The Code of Practice focuses on making AI-generated content identifiable. Its guidance covers both labeling and detection, with particular attention to deepfakes. It also refers to EU icons that can be used to label AI-generated content, and to how labeling should be applied where content has been generated or manipulated in the public interest.

That distinction matters because AI output appears in several forms. A company might use a model to draft product copy, generate an image for a campaign, create synthetic audio, alter a video, or provide answers through a chatbot. These uses have different audience expectations and different potential for confusion. A realistic altered video, for example, presents a more direct transparency challenge than internal brainstorming material that is never published.

The code does not turn every business use of AI into the same implementation exercise. Instead, it provides a common framework for deployers to consider when AI-generated or manipulated material is made available to others. Businesses should use the code as practical guidance alongside their own legal assessment of the AI Act obligations relevant to their activity.

A practical labeling workflow for generative AI

A sensible starting point is to map the points at which AI-assisted output reaches customers, users, or the public. The goal is to make labeling a repeatable part of publishing and service processes, rather than a decision left to an individual employee at the last minute.

Useful implementation steps include:

• Identify public-facing AI outputs, including images, videos, audio, written content, and chatbot interactions.
• Separate generation from material manipulation, especially where an existing image, recording, or video could lead viewers to believe something happened when it did not.
• Define an approval process for deciding when and how the relevant label or icon should appear.
• Build labels into templates and workflows so teams do not need to recreate disclosure practices for every campaign or customer interaction.
• Review supplier and platform settings to understand what labeling or detection information AI tools can support.

For chatbots, clear identification can help set appropriate expectations before a customer begins an interaction. For content teams, a consistent publication checklist can reduce the risk that AI-generated or manipulated assets are treated differently across websites, social channels, and campaigns.

The practical benefit is clarity. Transparent use of AI can reduce avoidable customer confusion while helping teams apply a consistent standard as generative tools become part of everyday work.

Online safety and marketplace design remain connected policy themes

The labeling code was published alongside other EU activity in August 2026 that points to a wider focus on online safety, consumer protection, and digital services. The Better Internet for Kids network's #NotOnOurFeed campaign runs from August 24 to September 25, 2026. It promotes protection, prevention, reporting, and support in response to cyberbullying, including AI-related abuse within its scope.

For operators of online communities, marketplaces, or customer platforms, this is a reminder that transparency and safety are connected. A labeling practice tells users what they are seeing or interacting with. Reporting and support processes give them a route to raise concerns when content or behavior causes harm.

Separately, the Commission has issued a call for tenders for a study examining how online marketplace design shapes user behavior. The work is specifically connected to the Digital Services Act and Very Large Online Marketplaces. The tender is not a new marketplace rule, and it does not establish conclusions about particular design choices. It does show that the Commission is seeking evidence on how platform interfaces and commercial design can influence users.

Smaller marketplace operators are not the stated focus of that study. Still, the direction of travel is relevant: product listings, recommendation flows, disclosures, complaint paths, and advertising presentation are not merely design details when they shape consumer decisions. Teams that document why users see content, promotions, or AI-generated interactions will be better placed to adapt if expectations evolve.

Businesses that want to use generative AI without creating inconsistent customer experiences need more than a one-off label. Scalevise can help translate AI use cases into workable processes, from mapping public-facing outputs to selecting practical implementation priorities through AI consultancy for practical adoption. A clear approach can reduce manual uncertainty and help teams introduce AI features with greater confidence. Request an AI consultancy conversation with Scalevise.

Frequently Asked Questions

What is the EU Code of Practice on marking and labeling AI-generated content?

It is European Commission guidance designed to support compliance with AI Act transparency obligations. It covers the labeling and detection of AI-generated content and deepfakes, including guidance for deployers of generative AI.

Does the code apply only to deepfakes?

No. Deepfakes are a major focus, but the code addresses AI-generated content more broadly and provides guidance on marking and labeling that content.

How can a business start implementing AI content labeling?

Start by identifying public-facing AI outputs, such as generated images, video, audio, written material, and chatbot interactions. Then establish a consistent process for deciding when and how labels should be displayed.

What is the #NotOnOurFeed campaign?

NotOnOurFeed is a Better Internet for Kids network cyberbullying awareness campaign running from August 24 to September 25, 2026. It emphasizes protection, prevention, reporting, and support, including in relation to AI-related abuse.

Does the DSA marketplace-design study create new obligations for online marketplaces?

No. The Commission has issued a call for tenders for a study on how marketplace design influences user behavior in the context of the DSA and Very Large Online Marketplaces. The study tender itself does not create new obligations.

Conclusion

The Commission's AI content labeling code gives generative AI deployers a concrete reference for making AI-generated and manipulated material more transparent. Alongside the EU's cyberbullying campaign and marketplace-design research, it reflects continued attention to how digital services inform, influence, and protect users. Businesses using AI in public-facing workflows should treat transparent labeling as a practical design and operating consideration, not an afterthought.