Might your AI content
need a label?
The transparency obligations of the EU AI Act have applied since 2 August 2026. Alongside them, every platform has its own rules, and they differ in the details. The usual result: people either label everything out of caution, or nothing at all.
Answer a few questions and get orientation on which labelling rules may be relevant for your content — separated into the AI Act, the platform rule and voluntary recommendations, each with an official source.
The checker does not replace legal advice and does not assess an individual legal case.
Why this matters right now
The transparency obligations in Article 50 of the AI Act have applied since 2 August 2026. In Germany the AI Market Surveillance and Innovation Promotion Act has been in force since 29 July 2026. Under it the Bundesnetzagentur is generally the central market surveillance authority and central complaints body; special competences may exist for individual regulated sectors. Separate from all of that are the platforms' own rules, which apply in their own right.
The platform rules at a glance
| Platform | What is covered | How it is labelled |
|---|---|---|
| TikTok | Realistic-looking AI-generated or materially AI-altered images, video and audio — including invented but realistic-seeming people, places or events. Obviously stylised or unrealistic content is as a rule outside the mandatory label; other TikTok rules can still apply. | A switch in the post settings before publishing. Beyond that, TikTok recommends transparency for fully AI-generated or materially AI-edited content. |
| Instagram / Facebook | Meta uses automatic and user-declared transparency mechanisms for AI content. For certain photorealistic video and realistic-sounding audio, your own disclosure may be required. | Depending on the content, declare it at upload; in addition, "AI info" notes can appear automatically from technical provenance signals. |
| No general mandatory AI label for every upload has been published. For certain synthetic or manipulated media LinkedIn requires clear disclosure, in particular where real people or events are depicted deceptively. | No general toggle. LinkedIn recommends transparency for heavily AI-assisted content and can display existing C2PA Content Credentials. | |
| Text only (all platforms) | Not covered by the platforms' labelling rules, which attach to image, video and audio. | As a rule no platform label is apparent. Statutory special cases under Article 50(4) of the AI Act are unaffected. |
Frequently asked questions
Do I have to label a caption written with AI?
The labelling rules of Instagram, Facebook, TikTok and LinkedIn attach to AI-generated image, video and audio content, not to text alone. The statutory layer is separate: Article 50(4) of the AI Act provides for disclosure of AI-generated text published to inform the public on matters of public interest. Ordinary product, brand and social media captions typically fall outside that; for political, journalistic or socially relevant text the provision may well be relevant.
Does human review change the assessment?
Article 50(4) of the AI Act carves out cases where the content underwent substantial human review or editorial control and a person or organisation holds editorial responsibility for it. What is meant is a review of the substance by a person with relevant knowledge — spell checking, grammar checking, skimming or a purely formal check are, per the European Commission's explanations, expressly not enough. Whether a review counts as sufficient in a specific case is a legal assessment this tool does not make.
Does the AI Act apply to private use?
The deployer obligations described here concern professional and business use in particular. Purely personal, non-professional use by natural persons may fall outside the scope of these deployer obligations. The rules of the individual platform apply regardless.
Since when do the transparency duties apply?
The transparency obligations in Article 50 of the AI Act have applied since 2 August 2026. Content created before that date does not have to be labelled retroactively under the Regulation.
Who is responsible in Germany?
The Bundesnetzagentur is generally the central market surveillance authority and central complaints body under the German AI Market Surveillance and Innovation Promotion Act, in force since 29 July 2026. Special competences may exist for individual regulated sectors.
What about filters and retouching?
Cropping, exposure, colour correction, light retouching and upscaling without changing the content are generally not treated as AI-generated content. Where an AI filter noticeably changes a face, a body or other essential features, however, that can count as substantial AI editing.
Does this tool replace a legal review?
No. It reproduces what the platforms have published and what the cited provisions address. It does not assess a specific case, gives no legal advice and promises no legal certainty. For commercially or legally sensitive publications, competent legal advice should be sought when in doubt.
By the way: most captions need no label
The platforms' labelling rules attach to image, video and audio, not to text. Ordinary product, brand and social media captions also typically fall outside the specific transparency duty for AI-generated text on matters of public interest. For political, journalistic or socially relevant text, Article 50(4) of the AI Act may well be relevant.
Join the Waitlist