-2.png?w=1024)
Europe’s AI labels are no longer optional. Companies must now tell users when they are interacting with certain AI systems and disclose when covered content has been generated or manipulated by AI.
New transparency obligations under the EU AI Act took effect on August 2, creating separate requirements for companies that provide AI systems and those that deploy them. The rules cover chatbots, deepfakes, biometric tools, and some AI-generated public-interest content.
The EU’s new AI disclosure requirements
The European Commission said companies must clearly inform users when they are interacting with an AI system instead of a real person. The obligation can apply to chatbots, AI agents, avatars, and other interactive systems unless their artificial nature is already obvious.
Providers, meaning companies that develop and market AI systems, must design those systems to support the required disclosures. Deployers, including platforms and services that use AI systems, are responsible for informing users and labeling covered content in the situations outlined by the rules.
Some companies may operate as both providers and deployers, The Verge reported, which could leave them responsible for obligations on both sides of the AI supply chain.
Which AI systems and content are covered
The rules apply to AI-generated or manipulated images, audio, and video that resemble real people, objects, places, entities, or events and falsely appear authentic or truthful.
Companies must also disclose the use of emotion recognition and biometric categorisation systems.
AI-generated text published to inform the public about matters of public interest must be labelled when it has not undergone human review or editorial control, according to the Commission. The requirement is intended to help people assess whether information came from a human-led editorial process or an automated system.
How labels and machine-readable marks work
Covered content must include a clear and visible disclosure as well as machine-readable marks that allow systems to detect that it was artificially generated or manipulated.
The Commission has created a set of AI disclosure icons that companies may adopt. The icons are optional, but the labeling obligations are mandatory.
Watermarks, metadata, and other digital markers may help companies meet the technical requirements. Platforms including Meta and TikTok have already introduced their own AI labels, although differences between platforms could make disclosures less consistent for users.
Euronews reported that Google had warned overlapping labels and legal notices could create confusion rather than provide clearer context.
Deadlines, exemptions, and potential fines
The transparency obligations apply immediately to new AI systems introduced after August 2. Systems and services released before that date have until December 2, 2026, to comply.
The rules generally apply to professional uses of AI rather than individuals using the technology solely for personal purposes. Euronews also noted exemptions for some artistic, creative, satirical, and fictional works.
Enforcement falls to national market surveillance authorities, the European AI Office, and the European Data Protection Supervisor when EU institutions are involved.
Companies that fail to comply may face fines of up to €15 million or 3% of global annual turnover. EU institutions, bodies, and agencies may face penalties of up to €750,000. The Commission said authorities should consider proportionality when imposing sanctions on small and medium-sized businesses and small mid-cap companies.
How companies can prepare for compliance
Businesses should first determine whether they qualify as an AI provider, deployer, or both. They can then identify which customer interactions, publishing workflows, and synthetic-media tools fall within the rules.
Companies using third-party AI platforms should also clarify which disclosure features the vendor provides and which responsibilities remain with the business using the system. Existing tools may require updates to user notices, content labels, metadata, or editorial review processes before the December deadline.
The rules may not eliminate deceptive AI content, but they raise the cost of leaving users in the dark. For companies operating in Europe, transparency is now part of deploying AI, not a feature that can be added later.
The requirements build on the EU’s June roadmap for consistent deepfake and AI-content disclosures. For more background, read EU Sets Rules Roadmap for Deepfake, AI Content Labels.