A practical Article 50 workflow for teams publishing AI-generated and AI-manipulated content: determine the duty, apply the right disclosure, and retain a defensible record.
Label Placement
Duty / Type
Theme
Audience Locale
Built for the people accountable when AI content goes public.
Compliance operations, not just labels
Replace ad-hoc wording, scattered approvals, and uncertain hand-offs with a workflow your team can repeat and explain.
Map role, modality, audience, and exemptions to the relevant Article 50 duty.
Reasoned result
Apply legible labels and localized wording suited to the media and publishing context.
24 EU languages
Keep the decision, disclosure, and technical record together for internal or regulatory review.
Defensible evidence
Structured assessment
Answer a short set of questions about your role, media, and publishing context.
Select your target audience location or distribution market.
Regulatory context
Since 2 August 2026, publishing without a defined transparency process is a current compliance exposure. A defensible programme starts by knowing which duty applies before content is released.
Built for European audiences
Use localized disclosure wording based on official EU terminology, with consistent treatment across campaigns, markets, and media types.
Review the disclosure guide →Practical analysis, technical breakdowns, and step-by-step guides on EU AI Act compliance.
As of August 2026, Article 50 transparency mandates are legally active across the EU. Discover who must comply, the legal differences between AI Providers and Deployers, how to label synthetic media, and how to avoid €15M fines.
Learn the practical rules for labeling AI content under EU AI Act Article 50. From visible corner badges and spoken audio preambles to JSON-LD text metadata and 24-language translation rules.
An engineering deep-dive into C2PA cryptographic manifests, JSON-LD schemas, social media metadata stripping vulnerabilities, and multi-layer defence-in-depth watermarking for Article 50(2) compliance.
If you generate or manipulate image, audio, video or text content with AI and publish it in the EU, Article 50 of the AI Act very likely applies to you — as a provider (machine-readable marking) or a deployer (visible disclosure for deepfakes and public-interest text). The obligation has applied since 2 August 2026, with penalties up to €15M or 3% of worldwide turnover.
Yes. The image label tool and the text disclosure guide are free, without an account. We are building a paid platform for teams that need bulk marking, C2PA signing, invisible watermarks and audit-ready evidence records — this free site is how we earn your trust first.
No. Everything happens locally in your browser using the canvas API. Your images never leave your device — you can even use the tool offline once the page has loaded.
Not always. Providers must also mark outputs in a machine-readable format (our PNG downloads embed a machine-readable disclosure record), and full compliance may involve C2PA Content Credentials, invisible watermarking and evidence records. Treat this as a solid first step, not a complete programme.
Run a structured assessment before your next AI asset is published.