AI Content Labeling in Zen: New Moderation Rules

Why AI Regulation is Important Right Now

In the summer of 2026, many myths spread in the information space about the mandatory labeling of content created by neural networks, under threat of fines. In reality, AI content labeling in Russia is structured differently – and Zen authors should understand the real norms, not just headlines. Let's break down what the new law actually requires, whom it concerns, and how it relates to running a channel on Zen.

AI Law 2026 Labeling: What Was Actually Adopted

The AI law 2026 labeling refers to Federal Law No. 243-FZ "On Supporting the Development of Artificial Intelligence Technologies in the Russian Federation," signed on July 26, 2026. The document introduces basic concepts in the field of AI, defines the status of sovereign and national models, and specifically establishes a norm for the possibility of content labeling. The main part of the law comes into force on September 1, 2026, while the provisions specifically concerning labeling will take effect later – from March 1, 2027.

It's important to understand: these are not amendments to the law "On Information," as sometimes mistakenly presented, but an independent regulatory act. The law only applies to audio and visual materials created using large fundamental AI models – article texts, posts, and newsletters do not fall under this norm at all.

Voluntary Labeling of Neural Network Content: A Right, Not an Obligation

The main source of confusion is the wording of the norm. Voluntary labeling of neural network content is enshrined as a user's right, not an obligation. Article 9 of the law states that an individual or company "is provided with the opportunity to place" a warning about the use of AI – meaning the author decides whether to put a label or not.

The obligation in this structure is placed not on authors, but on platforms. Federal Law 243 on AI labeling explicitly states: owners of websites, applications, and social networks with a daily audience of more than 500 thousand users must provide users with a technical tool for such labeling. The format of the labeling is determined by agreement between the platform and the user, not by a single state standard.

Requirements for AI Content Platforms: Whom the Norm Concerns

Requirements for AI content platforms apply to any service with personal pages and a sufficiently large audience – regardless of the content's subject matter. Given Zen's monthly audience of tens of millions of users, the platform formally falls into the category obligated to provide a technical capability for labeling when the relevant norms come into force in March 2027.

Platforms are not obliged to independently identify generated content – the obligation is limited precisely to providing a tool for voluntary labeling by the author. How to label AI content in Zen in practice will depend on the interface the platform implements by the time the norm comes into force – the specific mechanism has not yet been detailed in either the law or subordinate acts.

Fines for AI Content Without Labeling: Debunking the Myth

Fines for AI content without labeling are not provided for authors by law at all. Article 11 refers to general legislation without specific sanctions, and no amendments to the Code of Administrative Offenses have been adopted in this regard. The only responsibility in this area threatens platforms that fail to provide the technical capability for labeling – but even here, details are delegated to subordinate acts that have not yet been developed.

A separate bill, which introduced mandatory labeling with fines for legal entities, was introduced in November 2025 as amendments to the law "On Information," but the State Duma rejected it on June 10, 2026, due to uncertainty in technical feasibility. Industry experts noted that modern detectors are not yet able to reliably determine whether material was created by a neural network without errors.

Deepfakes Law Russia: A Separate Story

The deepfakes law in Russia regulates deepfakes differently than the labeling of ordinary AI content. The law adopted in July 2026 is framework in nature and does not introduce criminal liability for using neural networks to create substitute materials. A separate bill on the criminalization of illegal processing of personal data for creating deepfakes has been under consideration since January 2026 but has not yet been adopted.

This means that the fight against malicious deepfakes today is conducted within the framework of existing norms – on defamation, fraud, and personal data protection – and not through a specialized labeling mechanism. Authors who honestly use generative tools to create content are not threatened by any sanctions under these norms.

Machine-Readable AI Labeling: Why This Is Not About Zen

Machine-readable AI labeling, in the sense of technical watermarks and digital tags at the metadata level, remains outside Russian regulation for now – unlike, for example, the approach of the European Union, where Article 50 of the Artificial Intelligence Regulation provides for precisely such mandatory marking. The Russian law does not require machine-readable labeling and leaves the format of the mark to an agreement between the platform and the user.

Large foreign platforms have already begun to implement automatic recognition and labeling of generative content on their own initiative, without reference to local legislation. Generative content rules in Zen are currently being formed based on the platform's general requirements for the reliability and originality of materials, rather than specialized AI labeling – until the relevant norms come into force in 2027.