€99
6 files
Article 50
Labeling Kit
Article 50 content labeling without the guesswork
A labeling guide by medium, ready-made disclosure wording, deepfake notice templates, and a publisher workflow checklist. Label your AI-generated content correctly.
What's inside
Article 50 labeling guide
12-page guide on what must be labeled by medium (text, image, audio, video), exemptions, and timing
Disclosure wording library
Ready wording per medium and placement: caption, footer, metadata, spoken notice
Deepfake/synthetic media notice templates
Notices for realistic media, avatar spokespersons, and cloned voices
Publisher workflow checklist
2-page pre-publish QA: is it generated? What label type? Where does it go? Log it
Labeling decision tree
1-page flowchart from content type to required action
Labeling record log
Track each asset: date, medium, label applied, reviewer
Who needs this
- ✓ You publish AI-generated text, images, audio, or video
- ✓ You use avatars, synthetic voices, or deepfake-adjacent media
- ✓ You need a repeatable labeling workflow, not one-off decisions
Who doesn't
- ✗ You don't publish any AI-generated content externally
- ✗ You already have a documented labeling process with records
FAQ
Does all AI-assisted text need to be labeled?
No. The text obligation is scoped to publications on matters of public interest, with an exception when a human exercises editorial review and takes responsibility. Other media have their own rules under Article 50.
What about images and videos?
Providers must ensure AI-generated outputs are marked machine-readable and detectable. Deployers of deepfakes must disclose that content is artificially generated.
Do we need to label internal documents?
Article 50 focuses on content published or shared with the public. Internal documents are generally outside scope, though your AI-use policy should still address them.
Example wording. Adapt to your context. Not legal advice.