EU AI Act: обязательная маркировка ИИ-контента и чат-ботов вступила в силу

The European Union on August 2, 2026, brought into force transparency obligations under the AI Act: companies are now required to disclose the fact of users interacting with AI systems and to label content generated or modified by neural networks.

What has changed for companies?

The new requirements distinguish responsibility between two types of market participants. Providers — companies that develop and bring AI systems to market — are obliged to embed disclosure mechanisms directly into the architecture of their products. Deployers — platforms and services using ready-made AI systems — must explicitly notify end users of their interaction with AI.

The boundary between the two statuses is not always clear: a number of major players, in particular Meta and SpaceXAI, simultaneously develop their own AI systems and integrate them into their platforms — meaning they bear responsibility both as providers and as deployers.

The distinction between categories is fundamental: the provider is responsible for ensuring the system is capable of disclosing its nature; the deployer is responsible for ensuring that this disclosure actually takes place at the point of contact with the user.

What exactly needs to be labeled?

Chatbots, deepfakes, and any synthetic content fall under the new requirements. According to the European Commission, the key parameters are as follows:

• Effective date — August 2, 2026
• Chatbots must explicitly inform the user that they are communicating with an AI, not a human
• Synthetic content — images, video, and audio generated or modified by AI — must be appropriately labeled
• Deepfakes are subject to the same disclosure requirements
• Companies combining both roles (such as Meta and SpaceXAI) bear cumulative obligations

The European Commission has developed a standard set of AI icons — a ready-made visual language that platforms can use instead of creating their own designations. The logic is simple: if enough services adopt the same icons, users will learn to recognize them more quickly.

"A safer and more transparent AI" — this is how the

European Commission framed the goal of the new package of requirements in its official press release of August 2, 2026.

Why does this matter beyond the EU?

The AI Act is de jure limited to Europe, but de facto affects any platforms operating with European users. For international services, this means specific compliance tasks: revising UX notifications, auditing AI feature interaction scenarios, and integrating standard labeling. According to European Commission materials, the AI Act is the world's first comprehensive legal instrument of this scale in the field of AI.

Non-compliance with the new requirements carries sanctions under the supervisory mechanisms that EU regulators are building in parallel with the Act itself. Similar initiatives on labeling AI-generated content are being discussed in the United States, the United Kingdom, and a number of Asian countries — so the practice introduced in Europe risks becoming a de facto global standard.

What this means

The EU AI Act introduces the first mandatory mass labeling regime for AI content in a major region. For users — this is minimal protection against hidden chatbots and unlabeled deepfakes. For the industry — a signal that regulatory pressure on AI transparency will only increase.

Frequently asked questions

When did the EU AI

Act transparency requirements come into force?

The transparency requirements came into force on August 2, 2026 — this date is enshrined in official European Commission documents.

What is the difference between provider and deployer obligations?

The provider — the company that created the AI system — is obliged to design it with mechanisms for explicit disclosure of its nature. The deployer — the platform using the system — is responsible for ensuring the user actually receives this disclosure at the moment of interaction. Companies like Meta and SpaceXAI are simultaneously providers and deployers and bear both types of obligations.

*Meta has been recognized as an extremist organization and is banned in Russia.