Deepfake disclosure EU AI Act
Deepfakes and synthetic media must be recognizable. Make your labeling demonstrable.
Article 50 divides duties by actor. Providers machine-mark synthetic output under Article 50(2). Under Article 50(4), deployers disclose when image, audio, or video is a deepfake; they also disclose certain AI-generated or manipulated text published to inform the public on matters of public interest. Only providers of Article 50(2) systems placed on the market before 2 August 2026 have until 2 December 2026 for machine-readable marking. LearnWize helps content teams apply that division of roles.
Marking duty
Deepfakes
Synthetic media
Evidence by asset
Evidence chain
LearnWize Article 4
1
Role2
Use case3
Risk4
Learning path5
Scenario6
Score7
Certificate8
ReportArticle 4
AI-generated media spreads faster than your labeling policy.
Marketing and communications generate images, voices, and video with AI. Without clear agreements on marking, labeling becomes inconsistent and it is hard to show afterwards that the duty was understood.
Generated visuals and video go live without marking.
Synthetic voice-overs are not made recognizable.
There is no consistent labeling approach per channel.
There is no evidence the team knows the duty.
Evidence
What LearnWize records around generated media
Role and AI use
Document which teams use AI, which tasks are involved, and which knowledge each role needs.
Risk and context
Connect training to the context in which AI is used, including impact on candidates, citizens, customers, or employees.
Learning paths and scenarios
Let employees practice with realistic work situations, not only generic AI explanations.
Testing and certificates
Keep participation, scores, certificates, and completion records as evidence that understanding was tested.
Management reporting
Give HR, Legal, Compliance, IT, and leadership a clear view of where the organization stands.
Refresh log
Decide when training needs to be updated because of new tools, policies, roles, or risks.
Teams
Typical participants
Creative and content teams
For generated images, video, audio, and design.
Social and brand
For campaigns, ads, and posts with synthetic media.
Communications and PR
For public content where recognizability matters most.
The 30-day route
1
Scan
Map roles, AI use, and missing evidence.
2
Classify
Connect teams to use cases, risks, and learning paths.
3
Activate
Launch LearnWize training by audience.
4
Lock
Test understanding with scenarios and certificates.
5
Evaluate
Deliver reporting, evidence dossier, and refresh advice.
Continue your route
Choose the variant that fits your question: training, course, masterclass, speaker, or evidence.
Article 50 transparency training
Article 50 transparency training
Article 50 applies from 2 August 2026. Train marketing, communications, and product teams on chatbot disclosure, deepfake labeling, and AI content transparency, with evidence per role.
Chatbot disclosure obligation
Chatbot disclosure obligation (Article 50)
Article 50 requires people to know when they are talking to a chatbot. Train the teams behind your chatbots and make the disclosure demonstrable.
AI content transparency
AI content transparency (Article 50)
Determine by actor and use case which Article 50 duty applies to chatbots, machine-readable marking, deepfakes, public-interest text, and certain biometric uses.
Official sources
Check the legal basis
For legal status and wording, we refer to primary official sources. Check both the applicable base text and Regulation (EU) 2026/1744, plus any later amendments.
EU AI Act, Regulation (EU) 2024/1689
The official original base text of the EU AI Act. For the current Article 4 wording and timelines, read it together with Regulation (EU) 2026/1744.
Regulation (EU) 2026/1744, Digital Omnibus on AI
The official amending regulation in force, including the new Article 4 wording and revised AI Act timelines.
Frequently asked questions
Frequently asked questions about AI literacy
What transition applies to existing AI systems?
Only providers of Article 50(2) systems placed on the market before 2 August 2026 have until 2 December 2026 to comply with machine-readable marking. Deployer disclosure duties under Article 50(4) do not have that transition.
Is this a legal compliance guarantee?
No. LearnWize helps organizations organize AI literacy in a practical and demonstrable way. Formal legal advice or a compliance opinion should be assessed separately.
Why is a certificate alone not enough?
A certificate is useful, but often misses the link to role, AI use, risk context, scenario practice, and management reporting. That chain makes the evidence stronger.
Where do we start if we do not have a clear overview yet?
Start with the AI Literacy Readiness Scan. It maps roles, AI use, priorities, and missing evidence pieces.
Who delivers this
Three brands, one evidence dossier
Who delivers AI literacy you can actually prove? LearnWize is the training and evidence platform that delivers role-based learning paths, testing, and an audit-ready dossier. Embed AI runs the AI governance scan and the 30-day Readiness Sprint. Responsible AI Platform is the knowledge source that explains the EU AI Act and Article 4. Together you move from explanation to training to evidence.
LearnWize
Training and evidence platform: learning paths, testing, certificates, and an Article 4 evidence dossier per role.
Embed AI
AI governance scan (2,950 euros, creditable) and the 30-day Readiness Sprint (9,900 euros) as a baseline and starting point.
Responsible AI Platform
Knowledge source that explains what the EU AI Act and Article 4 require and how they relate to Annex III.