Article 50 transparency check: map the right duty by actor
The Article 50 transparency duties have applied since 2 August 2026. The precise duty depends on your role as provider or deployer and on the use case. In a short check we map chatbots, generative AI and synthetic content, then record which disclosure, machine-readable marking or information duty applies.
Transparency check
1 week
From AI use to demonstrable transparency
Overview of chatbots, generative AI and AI content
Per use case: which Article 50 duty applies
Disclosure, marking and information texts
Action list with owner and priority for the applicable Article 50 duties
Transparency has not been postponed
Regulation (EU) 2026/1744 moves most standalone Annex III duties to 2 December 2027 and has been in force since 27 July 2026. Article 50 has applied since 2 August 2026. That does not mean every AI item needs a label: first determine whether you are provider or deployer and which paragraph covers the concrete use case.
The four transparency obligations of Article 50
Article 50 is separate from the high-risk regime but applies only within four actor- and use-case-specific duties. We assess each use case separately.
Direct AI interaction by providers
Providers of systems intended to interact directly with natural persons inform them that they are interacting with AI, unless this is obvious from the circumstances.
Machine-readable marking by providers
Providers of systems that generate synthetic audio, image, video or text machine-mark output under Article 50(2) and its applicable conditions.
Emotion recognition and biometric categorisation by deployers
Deployers inform exposed persons when using emotion-recognition or biometric-categorisation systems, subject to the applicable rules.
Deepfakes and certain public-interest text by deployers
Deployers disclose deepfakes and certain AI-generated or manipulated public-interest text, taking account of the exceptions and modalities in paragraph 4.
Transition period: Regulation (EU) 2026/1744 gives providers of relevant Article 50(2) systems placed on the market before 2 August 2026 until 2 December 2026 for the machine-readable marking duty. That transition does not automatically extend to deployer duties under paragraphs 3 and 4.
When this check fits
This check is for organizations that need to determine which Article 50 duty applies to each chatbot, generative system or publication channel and who owns the action.
You use chatbots or AI assistants
Customer service bots, conversational AI or virtual assistants where it is unclear whether disclosure is handled correctly.
You publish synthetic content
Marketing, social, product imagery or public-interest text where provider and deployer duties have not yet been separated.
No high-risk, possible Article 50 duty
A concrete use case may still trigger a transparency duty outside the high-risk regime.
What the check delivers
Inventory of chatbots, generative AI and AI content within scope
Per use case: which Article 50 obligation applies
Role determination: whether you act as provider or deployer per use case
Gap overview: which disclosure, marking or information is still missing
Concrete disclosure and information texts for bots, content and interfaces
Approach for provider marking and, where paragraph 4 applies, deployer disclosure
Action list with owner and priority for the applicable duties
Note on the transition period for existing generative AI
Approach in 1 week
Scope and intake
We define which chatbots, generative AI use cases and AI content fall within the transparency scope.
Inventory
We map per use case how AI is deployed toward users, customers or the public.
Test against Article 50
We connect each use case to the right transparency duty and determine whether you are provider or deployer.
Texts and marking
We deliver draft disclosures and an approach for machine-readable marking within the established actor and use-case scope.
Action list
You get a concrete list of actions, owners and priorities for the applicable Article 50 duties.
Who this works for
Marketing and communications
Teams that use generative AI for content, imagery or social and want marking and recognizability handled properly.
Customer service and digital
Teams with chatbots, virtual assistants or conversational AI in customer channels.
Legal, privacy and compliance
Teams that want to translate the transparency duties of Article 50 into concrete texts and evidence.
Product and engineering
Teams that need to build disclosure and machine-readable marking into interfaces and output.
Afterwards you know
Which Article 50 duty applies per use case
Whether you are provider or deployer per use case
Which disclosure and marking is still missing
Which texts and interface changes are needed
Which actions to repair first, because the duty is already running
Logical next steps
AI Act readiness and gap analysis
For a broader view of AI systems, roles, risks and evidence in one roadmap.
View routeEU AI Act deadline 2026
For the full picture: which obligations already apply and which date comes next.
View routeAI inventory setup
For a compact inventory of AI systems with owner, purpose and evidence status.
View routeDeepfake compliance
Paragraph 4 in practice: making synthetic image, audio and video recognisable.
View routeAI voices in advertising
For agencies and advertisers working with voice clones and synthetic voice-over.
View routeChatbot transparency
For paragraph 1: disclosure on direct AI interaction, including AI agents.
View routeMachine-readable marking of AI content
For paragraph 2: the provider duty and the transition until 2 December 2026.
View routeEmotion recognition and biometrics
For paragraph 3, preceded by the question whether the use is permitted at all.
View routeCode of practice on AI content transparency
Sign or demonstrate yourself, with a gap analysis against the code.
View routeBackground and deadlines
Deeper analysis on Praxikon about the transparency deadline and the postponement of the high-risk regime:
Frequently asked questions
Is Article 50 really not postponed?
The transparency obligations of Article 50 have applied since 2 August 2026 and were not postponed. Regulation (EU) 2026/1744 moves the Annex III high-risk regime to 2 December 2027 but leaves the Article 50 start date in place. Transparency is therefore the next obligation that applies.
Does this apply if we have no high-risk AI?
It can. Article 50 is separate from the high-risk regime, but the precise duty depends on your role and use case. Not every chatbot, AI tool or publication automatically receives the same disclosure or marking duty.
What is the transition period for existing generative AI?
Regulation (EU) 2026/1744 gives providers of relevant Article 50(2) systems placed on the market before 2 August 2026 until 2 December 2026 for the machine-readable marking duty. That transition does not automatically extend to deployer duties under paragraphs 3 and 4.
Is this legal advice?
No. This is a practical transparency check. We structure use cases, duties and concrete texts so marketing, communications, legal, IT and management can act in a focused way. For formal legal advice, involve legal counsel.
How much internal time does this require?
Usually 1 to 2 short interviews, an overview of your chatbots and generative AI use cases and one review moment. We do most of the drafting.
Map your Article 50 duties by actor and use case.
Start with the transparency check. We determine which applications are in scope and which disclosure, marking or information duty fits your role.