{"content_id":"bsl2atbahn","slug":"eu-ai-act-article-50-transparency-rules-2026","locale":"en","schema_type":"Article","category":"ai_data","category_name":"AI Data","title":"EU AI Act Article 50 Labeling Rules for 2026","summary":"EU AI Act Article 50 divides chatbot disclosure and synthetic content labeling obligations between providers and deployers. This guide covers the rules applicable from August 2, 2026, the limited transition period, exceptions, and penalty structure.","sponsorship_disclosure":null,"affiliate_disclosure":null,"commerce_disclosure":null,"author":{"name":"Injoys Editorial Team","url":"https://injoys.com/ko/about"},"key_points":["Article 50's transparency obligations apply from August 2, 2026.","Providers of interactive AI must inform users that they are interacting with AI.","Providers of synthetic content must apply machine-readable markings to outputs.","Deployers of deepfakes and certain public-interest texts must visibly disclose the use of AI.","The transition period until December 2, 2026, for certain existing systems is not a general grace period that postpones all obligations."],"content_markdown":"Article 50 of the EU AI Act requires chatbot disclosures, machine-readable markings for synthetic content, and visible disclosures for deepfakes and public-interest text from August 2, 2026. Only certain existing systems are granted a limited transition until December 2.\n\nAs of August 2026\n\n## What Changed on August 2, 2026\n\nThe transparency obligations under Article 50 apply from August 2, 2026. They cover AI providers and deployers that use AI professionally. The type of obligation varies depending on the system’s function and the purpose of the content.\n\nThe key changes fall into the following four categories.\n\n- Interaction disclosures informing users that they are communicating with AI\n- Machine-readable markings for synthetic images, audio, video, and text\n- Visible disclosures for deepfakes and certain public-interest text\n- Disclosures to people exposed to emotion recognition or biometric categorization systems\n\nArticle 50 is not a rule requiring the same wording to be attached to every AI output. Provider obligations and deployer obligations are distinct. Machine-readable markings and human-visible disclosures are also separate.\n\n## Comparison of Provider and Deployer Obligations\n\nProviders are responsible for transparency at the system level. Deployers are responsible for the context in which content is actually used. Whether one company may hold multiple roles should be checked against the relevant definitions and obligations in Regulation (EU) 2024/1689.\n\n| Category | Main obligation | Intended recipient of disclosure | Typical examples |\n|---|---|---|---|\n| Conversational AI provider | Disclose that the user is interacting with AI | User | Chatbot, AI customer service representative, AI agent |\n| Generative AI provider | Apply machine-readable markings to outputs | Distribution systems and detection tools | Synthetic images, audio, video, and text |\n| Deepfake deployer | Visibly disclose that content was artificially generated or manipulated | People viewing the content | Synthetic video that appears to depict a real person |\n| Public-interest text deployer | Disclose that content was generated or manipulated by AI | Readers of the text | Text addressing public affairs or social issues |\n| Emotion and biometric system deployer | Disclose that the system is operating | People exposed to the system | Users subject to emotion recognition or biometric categorization |\n\nHere, a provider is an entity that develops an AI system and places it on the market. A deployer is an entity that uses a system under its authority. Purely personal, non-professional activities are excluded from the definition of deployer.\n\n## Disclosure Standards for Chatbots and AI Agents\n\nAs a rule, conversational AI must disclose that it is AI. The timing of the disclosure must be checked in the text of Article 50 of Regulation (EU) 2024/1689. The point at which users must be informed should be checked in the text of Article 50 of Regulation (EU) 2024/1689.\n\nAn exception may apply, however, where it is obvious that the system is AI. This is assessed from the perspective of a reasonably well-informed, observant, and circumspect user. It is not sufficient that the developer internally classifies it as AI.\n\nAn exception may also apply where the law authorizes criminal investigations or similar activities. This exception operates only within the scope of the relevant legal authority. It does not automatically apply to ordinary customer service or marketing.\n\n## Machine-Readable Markings for AI-Generated Content\n\nGenerative AI providers must place detectable markings on synthetic outputs. Covered content includes images, audio, video, and text. A marking is a technical mechanism distinct from a human-visible label.\n\nMarkings must take into account effectiveness, interoperability, robustness, and reliability. Technical feasibility and implementation costs must also be considered. The generally acknowledged state of the art at the time is another factor.\n\nExceptions may apply to functions that assist with standard editing. Functions that do not substantially alter the meaning of the input may also qualify. The boundary between simple correction and the generation of new meaning should be documented.\n\n## Labeling Deepfakes and Public-Interest Text\n\nDeepfake deployers must disclose the manipulation in a way that people can see. A deepfake is synthetic content made to appear as though it depicts a real person, object, place, entity, or event. The definition of deepfake in Article 50 focuses on images, audio, and video.\n\nThe disclosure obligation does not disappear for artistic, creative, satirical, or fictional works. However, disclosure in a manner that does not hamper the display or enjoyment of the work is permitted. The location and format of the disclosure may be adapted to the context of the content.\n\nAI-generated or manipulated text published to inform the public on matters of public interest is also subject to disclosure. An exception may apply where there has been human review or editorial control. The natural or legal person holding editorial responsibility for publication must also be clearly identified.\n\nThe mere fact that a person clicked the publish button is not sufficient. It must be possible to demonstrate an actual review process and editorial responsibility. Review records provide a basis for determining whether an exception applies.\n\n## Summary by Condition\n\nWhether a disclosure is required must be assessed by considering both the content format and the context of use. The following table provides a quick distinction between the boundaries of Article 50. Individual cases should be assessed based on the actual functionality and distribution method.\n\n| Condition | General treatment | Issues to verify |\n|---|---|---|\n| Chatbot that users cannot clearly recognize as AI | Provide disclosure at the first interaction | Placement and accessibility of the disclosure |\n| AI generates a new image, audio, video, or text | Apply a machine-readable marking | Persistence and detectability of the marking |\n| Performs only standard editing of color, noise, or similar elements | Possible exception under Article 50(2) | Whether the meaning of the input was substantially altered |\n| Publicly releases synthetic video that appears to depict a real person | Apply a visible deepfake label | Whether appropriate disclosure is provided even for a creative work |\n| Publishes unedited AI text concerning matters of public interest | Apply a visible AI-use label | Whether there was human review and editorial responsibility |\n| Used by an individual for non-professional purposes | Possible exclusion from deployer obligations | Whether the use has shifted to business or professional activity |\n| Created outside the EU with outputs used in the EU | May be covered | Placement on the EU market and where the outputs are used |\n\nCompanies outside the EU are not automatically excluded. The rules may apply if they place a system on the EU market. Outputs from operators in third countries may also be subject to review if they are used in the EU.\n\n## Comparison of Disclosure Methods\n\nMachine-readable markings and visible disclosures serve different purposes. Applying one does not satisfy the other obligation. The same content may require both methods.\n\n| Disclosure method | Main audience | Purpose | Implementation examples |\n|---|---|---|---|\n| Machine-readable marking | Platforms and detection systems | Automatically identify whether content is synthetic | Metadata, watermarks, provenance information |\n| Visible disclosure | General users | Directly communicate that content has been manipulated | On-screen wording, audio notice, content description |\n| Interaction disclosure | Chatbot users | Communicate that the other party is AI | Notice before the conversation begins, disclosure in the first response |\n\nNo single specific technology is fixed in the legal text. Markings must remain robust after transformation and redistribution. An appropriate combination should be designed for the characteristics of the service.\n\n## Example of Calculating the Transition Period for Existing Systems\n\nThe limited transition period is not a general grace period for all existing services. It may apply only until December 2, 2026 to certain existing systems covered by Article 50(2). The precise scope must be checked in the European Commission’s guidelines.\n\n1. Check whether the system was placed on the market before August 2, 2026.\n2. Determine whether the obligation concerns the machine-readable marking under Article 50(2).\n3. Compare the system against the eligibility conditions for the limited transition in the EU guidelines.\n4. If eligible, complete the technical measures by December 2, 2026.\n5. Apply chatbot disclosures and deployer disclosure obligations separately.\n\nFor example, an existing generative system may be combined with a chatbot. A limited transition may be considered for markings on generated content. The application date for chatbot disclosures must be checked in Article 50 of Regulation (EU) 2024/1689 and the provisions concerning application and transition.\n\n## Common Mistakes\n\nThe most common misunderstanding is assuming that adding a single statement to AI content is sufficient. In practice, provider and deployer obligations are separate. The output format and disclosure context must also be assessed separately.\n\n- Mistaking a machine-readable marking for a visible deepfake label\n- Assuming that all existing systems have a grace period until December 2\n- Treating business promotion as personal use merely because it is conducted through a personal account\n- Assuming editorial control merely because a person uploaded the content\n- Treating every simple AI assistance function as covered by the standard-editing exception\n- Assuming that deepfake disclosure is waived for artistic works\n\nAccessibility requirements are also easy to overlook. Disclosures must be provided in a clear and distinguishable manner. Users with disabilities must also be able to perceive them.\n\n## Designing Evidence Across Distribution Stages\n\nDisclosure obligations must be managed not only when content is generated but also throughout the distribution process. Technical markings may disappear through compression, conversion, or re-uploading. Visible wording may also be omitted from the distribution screen.\n\nIt is safer for operational records to link the following items.\n\n- Model and system version used\n- Type of content generated or manipulated\n- Method used to apply the machine-readable marking\n- Visible disclosure wording and timing of exposure\n- Human reviewer and entity holding editorial responsibility\n- Tests confirming that markings are retained during transformation and distribution\n\nThese records are not a separate certificate required by law. However, they provide evidence explaining how the obligations were fulfilled. Responsibility for retaining disclosures may also be allocated in supply contracts.\n\n## Supervisory Authorities and the Reporting and Penalty Structure\n\nThe authorities responsible for enforcing Article 50 should be identified by checking the enforcement provisions of Regulation (EU) 2024/1689. Member States’ market surveillance authorities and competent authorities conduct on-site supervision. Authority over providers of general-purpose AI models may lie with the European Commission and the AI Office.\n\nAffected persons may lodge complaints with a market surveillance authority. The authority that receives a complaint depends on each Member State’s designated structure. The official AI supervisory contact point in the country where the service was provided should be checked.\n\nViolations of Article 50 may fall within the penalty category for breaches of other obligations under the EU AI Act. The specific penalty ceiling must be checked in the penalty provisions of Regulation (EU) 2024/1689. The actual amount is determined by taking into account factors such as the nature of the violation and the size of the company.\n\nSeparate ceiling principles apply to small and medium-sized enterprises. Fines are not imposed as an immediate fixed amount upon a violation. Member State procedures and proportionality assessments also apply.\n\n## Where to Check the Legal Text and Detailed Guidelines\n\nThe original legal obligations are set out in Article 50 of Regulation (EU) 2024/1689. Detailed interpretations should be checked in the European Commission’s Article 50 transparency guidelines. The official FAQ supplements examples involving chatbots, deepfakes, and public-interest text.\n\nCodes of practice are tools for aligning technical markings and labeling methods. Participation in a voluntary code does not eliminate the legal obligations themselves. The legal text and enforcement guidelines remain the primary criteria for assessment.\n\n## Frequently Asked Questions\n\n### Must AI chatbots always be labeled as AI?\n\nAs a rule, users must be informed that they are interacting with AI. However, an exception may apply if it is obvious from the circumstances that the system is AI. The timing of the disclosure must be checked in the text of Article 50 of Regulation (EU) 2024/1689.\n\n### Does a machine-readable marking alone also satisfy the deepfake disclosure requirement?\n\nThe two obligations are mechanisms intended for different recipients. A technical marking assists machine detection. A deepfake disclosure must inform people that the content has been manipulated.\n\n### Must every text written by AI be labeled?\n\nNot all AI-written text is subject to deployer disclosure. The primary target is text published to inform the public on matters of public interest. An exception may apply where there is human review and editorial responsibility.\n\n### Must Korean companies also comply with Article 50?\n\nThe rules may apply if they provide AI systems in the EU market. Outputs created in a third country may also be covered if they are used in the EU. Applicability cannot be determined solely by the location of the legal entity.\n\n### Are all obligations postponed until December 2, 2026?\n\nNo. The limited transition applies only to certain existing systems. It is not a period that uniformly postpones chatbot disclosures and deployer disclosure obligations.","content_html":"\u003cp\u003eArticle 50 of the EU AI Act requires chatbot disclosures, machine-readable markings for synthetic content, and visible disclosures for deepfakes and public-interest text from August 2, 2026. Only certain existing systems are granted a limited transition until December 2.\u003c/p\u003e\n\u003cp\u003eAs of August 2026\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#what-changed-on-august-2-2026\" class=\"anchor\" id=\"what-changed-on-august-2-2026\"\u003e\u003c/a\u003eWhat Changed on August 2, 2026\u003c/h2\u003e\n\u003cp\u003eThe transparency obligations under Article 50 apply from August 2, 2026. They cover AI providers and deployers that use AI professionally. The type of obligation varies depending on the system’s function and the purpose of the content.\u003c/p\u003e\n\u003cp\u003eThe key changes fall into the following four categories.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eInteraction disclosures informing users that they are communicating with AI\u003c/li\u003e\n\u003cli\u003eMachine-readable markings for synthetic images, audio, video, and text\u003c/li\u003e\n\u003cli\u003eVisible disclosures for deepfakes and certain public-interest text\u003c/li\u003e\n\u003cli\u003eDisclosures to people exposed to emotion recognition or biometric categorization systems\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eArticle 50 is not a rule requiring the same wording to be attached to every AI output. Provider obligations and deployer obligations are distinct. Machine-readable markings and human-visible disclosures are also separate.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#comparison-of-provider-and-deployer-obligations\" class=\"anchor\" id=\"comparison-of-provider-and-deployer-obligations\"\u003e\u003c/a\u003eComparison of Provider and Deployer Obligations\u003c/h2\u003e\n\u003cp\u003eProviders are responsible for transparency at the system level. Deployers are responsible for the context in which content is actually used. Whether one company may hold multiple roles should be checked against the relevant definitions and obligations in Regulation (EU) 2024/1689.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eCategory\u003c/th\u003e\n\u003cth\u003eMain obligation\u003c/th\u003e\n\u003cth\u003eIntended recipient of disclosure\u003c/th\u003e\n\u003cth\u003eTypical examples\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Category\"\u003eConversational AI provider\u003c/td\u003e\n\u003ctd data-label=\"Main obligation\"\u003eDisclose that the user is interacting with AI\u003c/td\u003e\n\u003ctd data-label=\"Intended recipient of disclosure\"\u003eUser\u003c/td\u003e\n\u003ctd data-label=\"Typical examples\"\u003eChatbot, AI customer service representative, AI agent\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Category\"\u003eGenerative AI provider\u003c/td\u003e\n\u003ctd data-label=\"Main obligation\"\u003eApply machine-readable markings to outputs\u003c/td\u003e\n\u003ctd data-label=\"Intended recipient of disclosure\"\u003eDistribution systems and detection tools\u003c/td\u003e\n\u003ctd data-label=\"Typical examples\"\u003eSynthetic images, audio, video, and text\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Category\"\u003eDeepfake deployer\u003c/td\u003e\n\u003ctd data-label=\"Main obligation\"\u003eVisibly disclose that content was artificially generated or manipulated\u003c/td\u003e\n\u003ctd data-label=\"Intended recipient of disclosure\"\u003ePeople viewing the content\u003c/td\u003e\n\u003ctd data-label=\"Typical examples\"\u003eSynthetic video that appears to depict a real person\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Category\"\u003ePublic-interest text deployer\u003c/td\u003e\n\u003ctd data-label=\"Main obligation\"\u003eDisclose that content was generated or manipulated by AI\u003c/td\u003e\n\u003ctd data-label=\"Intended recipient of disclosure\"\u003eReaders of the text\u003c/td\u003e\n\u003ctd data-label=\"Typical examples\"\u003eText addressing public affairs or social issues\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Category\"\u003eEmotion and biometric system deployer\u003c/td\u003e\n\u003ctd data-label=\"Main obligation\"\u003eDisclose that the system is operating\u003c/td\u003e\n\u003ctd data-label=\"Intended recipient of disclosure\"\u003ePeople exposed to the system\u003c/td\u003e\n\u003ctd data-label=\"Typical examples\"\u003eUsers subject to emotion recognition or biometric categorization\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n\u003cp\u003eHere, a provider is an entity that develops an AI system and places it on the market. A deployer is an entity that uses a system under its authority. Purely personal, non-professional activities are excluded from the definition of deployer.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#disclosure-standards-for-chatbots-and-ai-agents\" class=\"anchor\" id=\"disclosure-standards-for-chatbots-and-ai-agents\"\u003e\u003c/a\u003eDisclosure Standards for Chatbots and AI Agents\u003c/h2\u003e\n\u003cp\u003eAs a rule, conversational AI must disclose that it is AI. The timing of the disclosure must be checked in the text of Article 50 of Regulation (EU) 2024/1689. The point at which users must be informed should be checked in the text of Article 50 of Regulation (EU) 2024/1689.\u003c/p\u003e\n\u003cp\u003eAn exception may apply, however, where it is obvious that the system is AI. This is assessed from the perspective of a reasonably well-informed, observant, and circumspect user. It is not sufficient that the developer internally classifies it as AI.\u003c/p\u003e\n\u003cp\u003eAn exception may also apply where the law authorizes criminal investigations or similar activities. This exception operates only within the scope of the relevant legal authority. It does not automatically apply to ordinary customer service or marketing.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#machine-readable-markings-for-ai-generated-content\" class=\"anchor\" id=\"machine-readable-markings-for-ai-generated-content\"\u003e\u003c/a\u003eMachine-Readable Markings for AI-Generated Content\u003c/h2\u003e\n\u003cp\u003eGenerative AI providers must place detectable markings on synthetic outputs. Covered content includes images, audio, video, and text. A marking is a technical mechanism distinct from a human-visible label.\u003c/p\u003e\n\u003cp\u003eMarkings must take into account effectiveness, interoperability, robustness, and reliability. Technical feasibility and implementation costs must also be considered. The generally acknowledged state of the art at the time is another factor.\u003c/p\u003e\n\u003cp\u003eExceptions may apply to functions that assist with standard editing. Functions that do not substantially alter the meaning of the input may also qualify. The boundary between simple correction and the generation of new meaning should be documented.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#labeling-deepfakes-and-public-interest-text\" class=\"anchor\" id=\"labeling-deepfakes-and-public-interest-text\"\u003e\u003c/a\u003eLabeling Deepfakes and Public-Interest Text\u003c/h2\u003e\n\u003cp\u003eDeepfake deployers must disclose the manipulation in a way that people can see. A deepfake is synthetic content made to appear as though it depicts a real person, object, place, entity, or event. The definition of deepfake in Article 50 focuses on images, audio, and video.\u003c/p\u003e\n\u003cp\u003eThe disclosure obligation does not disappear for artistic, creative, satirical, or fictional works. However, disclosure in a manner that does not hamper the display or enjoyment of the work is permitted. The location and format of the disclosure may be adapted to the context of the content.\u003c/p\u003e\n\u003cp\u003eAI-generated or manipulated text published to inform the public on matters of public interest is also subject to disclosure. An exception may apply where there has been human review or editorial control. The natural or legal person holding editorial responsibility for publication must also be clearly identified.\u003c/p\u003e\n\u003cp\u003eThe mere fact that a person clicked the publish button is not sufficient. It must be possible to demonstrate an actual review process and editorial responsibility. Review records provide a basis for determining whether an exception applies.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#summary-by-condition\" class=\"anchor\" id=\"summary-by-condition\"\u003e\u003c/a\u003eSummary by Condition\u003c/h2\u003e\n\u003cp\u003eWhether a disclosure is required must be assessed by considering both the content format and the context of use. The following table provides a quick distinction between the boundaries of Article 50. Individual cases should be assessed based on the actual functionality and distribution method.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eCondition\u003c/th\u003e\n\u003cth\u003eGeneral treatment\u003c/th\u003e\n\u003cth\u003eIssues to verify\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003eChatbot that users cannot clearly recognize as AI\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003eProvide disclosure at the first interaction\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003ePlacement and accessibility of the disclosure\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003eAI generates a new image, audio, video, or text\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003eApply a machine-readable marking\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003ePersistence and detectability of the marking\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003ePerforms only standard editing of color, noise, or similar elements\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003ePossible exception under Article 50(2)\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003eWhether the meaning of the input was substantially altered\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003ePublicly releases synthetic video that appears to depict a real person\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003eApply a visible deepfake label\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003eWhether appropriate disclosure is provided even for a creative work\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003ePublishes unedited AI text concerning matters of public interest\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003eApply a visible AI-use label\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003eWhether there was human review and editorial responsibility\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003eUsed by an individual for non-professional purposes\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003ePossible exclusion from deployer obligations\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003eWhether the use has shifted to business or professional activity\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Condition\"\u003eCreated outside the EU with outputs used in the EU\u003c/td\u003e\n\u003ctd data-label=\"General treatment\"\u003eMay be covered\u003c/td\u003e\n\u003ctd data-label=\"Issues to verify\"\u003ePlacement on the EU market and where the outputs are used\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n\u003cp\u003eCompanies outside the EU are not automatically excluded. The rules may apply if they place a system on the EU market. Outputs from operators in third countries may also be subject to review if they are used in the EU.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#comparison-of-disclosure-methods\" class=\"anchor\" id=\"comparison-of-disclosure-methods\"\u003e\u003c/a\u003eComparison of Disclosure Methods\u003c/h2\u003e\n\u003cp\u003eMachine-readable markings and visible disclosures serve different purposes. Applying one does not satisfy the other obligation. The same content may require both methods.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eDisclosure method\u003c/th\u003e\n\u003cth\u003eMain audience\u003c/th\u003e\n\u003cth\u003ePurpose\u003c/th\u003e\n\u003cth\u003eImplementation examples\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Disclosure method\"\u003eMachine-readable marking\u003c/td\u003e\n\u003ctd data-label=\"Main audience\"\u003ePlatforms and detection systems\u003c/td\u003e\n\u003ctd data-label=\"Purpose\"\u003eAutomatically identify whether content is synthetic\u003c/td\u003e\n\u003ctd data-label=\"Implementation examples\"\u003eMetadata, watermarks, provenance information\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Disclosure method\"\u003eVisible disclosure\u003c/td\u003e\n\u003ctd data-label=\"Main audience\"\u003eGeneral users\u003c/td\u003e\n\u003ctd data-label=\"Purpose\"\u003eDirectly communicate that content has been manipulated\u003c/td\u003e\n\u003ctd data-label=\"Implementation examples\"\u003eOn-screen wording, audio notice, content description\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Disclosure method\"\u003eInteraction disclosure\u003c/td\u003e\n\u003ctd data-label=\"Main audience\"\u003eChatbot users\u003c/td\u003e\n\u003ctd data-label=\"Purpose\"\u003eCommunicate that the other party is AI\u003c/td\u003e\n\u003ctd data-label=\"Implementation examples\"\u003eNotice before the conversation begins, disclosure in the first response\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n\u003cp\u003eNo single specific technology is fixed in the legal text. Markings must remain robust after transformation and redistribution. An appropriate combination should be designed for the characteristics of the service.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#example-of-calculating-the-transition-period-for-existing-systems\" class=\"anchor\" id=\"example-of-calculating-the-transition-period-for-existing-systems\"\u003e\u003c/a\u003eExample of Calculating the Transition Period for Existing Systems\u003c/h2\u003e\n\u003cp\u003eThe limited transition period is not a general grace period for all existing services. It may apply only until December 2, 2026 to certain existing systems covered by Article 50(2). The precise scope must be checked in the European Commission’s guidelines.\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003eCheck whether the system was placed on the market before August 2, 2026.\u003c/li\u003e\n\u003cli\u003eDetermine whether the obligation concerns the machine-readable marking under Article 50(2).\u003c/li\u003e\n\u003cli\u003eCompare the system against the eligibility conditions for the limited transition in the EU guidelines.\u003c/li\u003e\n\u003cli\u003eIf eligible, complete the technical measures by December 2, 2026.\u003c/li\u003e\n\u003cli\u003eApply chatbot disclosures and deployer disclosure obligations separately.\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eFor example, an existing generative system may be combined with a chatbot. A limited transition may be considered for markings on generated content. The application date for chatbot disclosures must be checked in Article 50 of Regulation (EU) 2024/1689 and the provisions concerning application and transition.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#common-mistakes\" class=\"anchor\" id=\"common-mistakes\"\u003e\u003c/a\u003eCommon Mistakes\u003c/h2\u003e\n\u003cp\u003eThe most common misunderstanding is assuming that adding a single statement to AI content is sufficient. In practice, provider and deployer obligations are separate. The output format and disclosure context must also be assessed separately.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eMistaking a machine-readable marking for a visible deepfake label\u003c/li\u003e\n\u003cli\u003eAssuming that all existing systems have a grace period until December 2\u003c/li\u003e\n\u003cli\u003eTreating business promotion as personal use merely because it is conducted through a personal account\u003c/li\u003e\n\u003cli\u003eAssuming editorial control merely because a person uploaded the content\u003c/li\u003e\n\u003cli\u003eTreating every simple AI assistance function as covered by the standard-editing exception\u003c/li\u003e\n\u003cli\u003eAssuming that deepfake disclosure is waived for artistic works\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eAccessibility requirements are also easy to overlook. Disclosures must be provided in a clear and distinguishable manner. Users with disabilities must also be able to perceive them.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#designing-evidence-across-distribution-stages\" class=\"anchor\" id=\"designing-evidence-across-distribution-stages\"\u003e\u003c/a\u003eDesigning Evidence Across Distribution Stages\u003c/h2\u003e\n\u003cp\u003eDisclosure obligations must be managed not only when content is generated but also throughout the distribution process. Technical markings may disappear through compression, conversion, or re-uploading. Visible wording may also be omitted from the distribution screen.\u003c/p\u003e\n\u003cp\u003eIt is safer for operational records to link the following items.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eModel and system version used\u003c/li\u003e\n\u003cli\u003eType of content generated or manipulated\u003c/li\u003e\n\u003cli\u003eMethod used to apply the machine-readable marking\u003c/li\u003e\n\u003cli\u003eVisible disclosure wording and timing of exposure\u003c/li\u003e\n\u003cli\u003eHuman reviewer and entity holding editorial responsibility\u003c/li\u003e\n\u003cli\u003eTests confirming that markings are retained during transformation and distribution\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThese records are not a separate certificate required by law. However, they provide evidence explaining how the obligations were fulfilled. Responsibility for retaining disclosures may also be allocated in supply contracts.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#supervisory-authorities-and-the-reporting-and-penalty-structure\" class=\"anchor\" id=\"supervisory-authorities-and-the-reporting-and-penalty-structure\"\u003e\u003c/a\u003eSupervisory Authorities and the Reporting and Penalty Structure\u003c/h2\u003e\n\u003cp\u003eThe authorities responsible for enforcing Article 50 should be identified by checking the enforcement provisions of Regulation (EU) 2024/1689. Member States’ market surveillance authorities and competent authorities conduct on-site supervision. Authority over providers of general-purpose AI models may lie with the European Commission and the AI Office.\u003c/p\u003e\n\u003cp\u003eAffected persons may lodge complaints with a market surveillance authority. The authority that receives a complaint depends on each Member State’s designated structure. The official AI supervisory contact point in the country where the service was provided should be checked.\u003c/p\u003e\n\u003cp\u003eViolations of Article 50 may fall within the penalty category for breaches of other obligations under the EU AI Act. The specific penalty ceiling must be checked in the penalty provisions of Regulation (EU) 2024/1689. The actual amount is determined by taking into account factors such as the nature of the violation and the size of the company.\u003c/p\u003e\n\u003cp\u003eSeparate ceiling principles apply to small and medium-sized enterprises. Fines are not imposed as an immediate fixed amount upon a violation. Member State procedures and proportionality assessments also apply.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#where-to-check-the-legal-text-and-detailed-guidelines\" class=\"anchor\" id=\"where-to-check-the-legal-text-and-detailed-guidelines\"\u003e\u003c/a\u003eWhere to Check the Legal Text and Detailed Guidelines\u003c/h2\u003e\n\u003cp\u003eThe original legal obligations are set out in Article 50 of Regulation (EU) 2024/1689. Detailed interpretations should be checked in the European Commission’s Article 50 transparency guidelines. The official FAQ supplements examples involving chatbots, deepfakes, and public-interest text.\u003c/p\u003e\n\u003cp\u003eCodes of practice are tools for aligning technical markings and labeling methods. Participation in a voluntary code does not eliminate the legal obligations themselves. The legal text and enforcement guidelines remain the primary criteria for assessment.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#frequently-asked-questions\" class=\"anchor\" id=\"frequently-asked-questions\"\u003e\u003c/a\u003eFrequently Asked Questions\u003c/h2\u003e\n\u003ch3\u003e\n\u003ca href=\"#must-ai-chatbots-always-be-labeled-as-ai\" class=\"anchor\" id=\"must-ai-chatbots-always-be-labeled-as-ai\"\u003e\u003c/a\u003eMust AI chatbots always be labeled as AI?\u003c/h3\u003e\n\u003cp\u003eAs a rule, users must be informed that they are interacting with AI. However, an exception may apply if it is obvious from the circumstances that the system is AI. The timing of the disclosure must be checked in the text of Article 50 of Regulation (EU) 2024/1689.\u003c/p\u003e\n\u003ch3\u003e\n\u003ca href=\"#does-a-machine-readable-marking-alone-also-satisfy-the-deepfake-disclosure-requirement\" class=\"anchor\" id=\"does-a-machine-readable-marking-alone-also-satisfy-the-deepfake-disclosure-requirement\"\u003e\u003c/a\u003eDoes a machine-readable marking alone also satisfy the deepfake disclosure requirement?\u003c/h3\u003e\n\u003cp\u003eThe two obligations are mechanisms intended for different recipients. A technical marking assists machine detection. A deepfake disclosure must inform people that the content has been manipulated.\u003c/p\u003e\n\u003ch3\u003e\n\u003ca href=\"#must-every-text-written-by-ai-be-labeled\" class=\"anchor\" id=\"must-every-text-written-by-ai-be-labeled\"\u003e\u003c/a\u003eMust every text written by AI be labeled?\u003c/h3\u003e\n\u003cp\u003eNot all AI-written text is subject to deployer disclosure. The primary target is text published to inform the public on matters of public interest. An exception may apply where there is human review and editorial responsibility.\u003c/p\u003e\n\u003ch3\u003e\n\u003ca href=\"#must-korean-companies-also-comply-with-article-50\" class=\"anchor\" id=\"must-korean-companies-also-comply-with-article-50\"\u003e\u003c/a\u003eMust Korean companies also comply with Article 50?\u003c/h3\u003e\n\u003cp\u003eThe rules may apply if they provide AI systems in the EU market. Outputs created in a third country may also be covered if they are used in the EU. Applicability cannot be determined solely by the location of the legal entity.\u003c/p\u003e\n\u003ch3\u003e\n\u003ca href=\"#are-all-obligations-postponed-until-december-2-2026\" class=\"anchor\" id=\"are-all-obligations-postponed-until-december-2-2026\"\u003e\u003c/a\u003eAre all obligations postponed until December 2, 2026?\u003c/h3\u003e\n\u003cp\u003eNo. The limited transition applies only to certain existing systems. It is not a period that uniformly postpones chatbot disclosures and deployer disclosure obligations.\u003c/p\u003e\n","tags":["EU AI Act","Generative AI","Personal data protection","AI Agents","Deepfake"],"faqs":[{"question":"When does Article 50 of the EU AI Act begin to apply?","answer":"The transparency obligations apply from August 2, 2026. A limited transition until December 2, 2026 may be allowed for machine-readable markings on certain existing systems."},{"question":"Must AI chatbots always disclose that they are AI?","answer":"The point at which the fact that it is AI must be disclosed should be verified in the original text of Article 50 of Regulation (EU) 2024/1689. An exception may be possible where it is obvious to a reasonably well-informed, observant, and circumspect user that it is AI, but the service must not make this determination arbitrarily."},{"question":"What markings are required for AI-generated content?","answer":"Providers of generative AI must apply machine-readable markings to synthetic images, audio, video, and text. Deployers that make deepfakes or certain public-interest texts publicly available may also be required to provide a separate disclosure that is recognizable by people."},{"question":"Is it sufficient to add only a machine-readable marking to a deepfake?","answer":"No. Machine-readable marking is an obligation on providers for automated detection. Deployers of deepfakes must also disclose that the content was artificially generated or manipulated in a way that people can see."},{"question":"Must deepfakes created for artistic or satirical purposes also be labeled?","answer":"The disclosure obligation does not disappear solely because the content is artistic, creative, satirical, or fictional. However, it may be labeled in an appropriate manner that does not hamper the display or enjoyment of the work."},{"question":"Are all news articles or public-interest texts written by AI subject to labeling requirements?","answer":"The primary targets are AI-generated or manipulated texts published to inform the public about matters of public interest. An exception may be possible if there has been substantial human review or editorial control and the party holding editorial responsibility is clearly identified."},{"question":"Is an AI-generated content marking also required for simple photo retouching?","answer":"Whether the function falls under an exception should be verified in the original text of Article 50 of Regulation (EU) 2024/1689. Work that creates new subjects or meaning is difficult to regard as simple editing."},{"question":"Does this also apply to AI content created by individuals as a hobby?","answer":"Purely personal, non-professional activities may be excluded from the definition of a deployer. However, if the activity extends to business promotion, paid work, or professional posting, the nature of the use must be reassessed."},{"question":"Does Article 50 also apply to AI companies outside the EU?","answer":"It may apply if they place systems on the EU market or provide services in the EU. The scope of application must also be reviewed where output generated outside the EU is used in the EU."},{"question":"How much is the fine for violating Article 50?","answer":"The specific maximum penalty for the category of violations of other obligations should be verified in the penalty provisions of Regulation (EU) 2024/1689. The actual penalty is determined by taking into account factors such as the nature and duration of the violation, the size of the company, and proportionality."}],"sources":[{"url":"https://eur-lex.europa.eu/eli/reg/2024/1689","title":"Regulation (EU) 2024/1689 — Artificial Intelligence Act","type":"source"},{"url":"https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act","title":"Transparency obligations under Article 50 of the AI Act","type":"source"},{"url":"https://digital-strategy.ec.europa.eu/en/news/commission-publishes-guidelines-transparency-obligations-providers-and-deployers-certain-ai-systems","title":"Commission guidelines on transparency obligations for providers and deployers","type":"source"},{"url":"https://digital-strategy.ec.europa.eu/en/news/commission-starts-enforcing-ai-act-rules-and-new-transparency-requirements-2-august","title":"Commission starts enforcing AI Act rules and new transparency requirements","type":"source"}],"images":[{"id":1100,"url":"https://injoys.com/rails/active_storage/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6MTU1MDYsInB1ciI6ImJsb2JfaWQifX0=--278d568892775304c127c609d86d7fc0478b4e29/ai-cfed75f9.webp","is_representative":true,"generation_method":"ai_photo","license":"ai_generated","mime_type":"image/webp","translations":{"ko":{"alt":"공항 터치스크린 키오스크의 가상 상담원과 상호작용하는 여성","caption":"여성이 화면 속 디지털 상담원이 표시된 셀프서비스 키오스크를 이용하고 있다.","description":null},"en":{"alt":"Woman interacting with a virtual assistant on an airport touchscreen kiosk","caption":"A woman uses a self-service kiosk displaying a digital assistant on screen.","description":null},"ja":{"alt":"空港のタッチスクリーン端末で仮想案内員を操作する女性","caption":"女性がデジタル案内員を表示したセルフサービス端末を利用している。","description":null},"es":{"alt":"Mujer interactuando con una asistente virtual en un quiosco táctil de aeropuerto","caption":"Una mujer utiliza un quiosco de autoservicio que muestra una asistente digital en pantalla.","description":null},"id":{"alt":"Perempuan berinteraksi dengan asisten virtual di kios layar sentuh bandara","caption":"Seorang perempuan menggunakan kios mandiri yang menampilkan asisten digital di layar.","description":null},"pt":{"alt":"Mulher interagindo com uma assistente virtual em um quiosque touchscreen de aeroporto","caption":"Uma mulher usa um quiosque de autoatendimento que exibe uma assistente digital na tela.","description":null},"zh-hant":{"alt":"女子在機場觸控資訊站與虛擬助理互動","caption":"一名女子正在使用螢幕上顯示數位助理的自助服務機台。","description":null},"de":{"alt":"Frau interagiert an einem Flughafen-Touchscreen-Terminal mit einer virtuellen Assistentin","caption":"Eine Frau nutzt ein Selbstbedienungsterminal, auf dessen Bildschirm eine digitale Assistentin erscheint.","description":null}}},{"id":1101,"url":"https://injoys.com/rails/active_storage/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6MTU1MTIsInB1ciI6ImJsb2JfaWQifX0=--ec7981404b7d0fdb6fc8b97fec21073d5e36fa0b/ai-5f39f330.webp","is_representative":false,"generation_method":"ai_image","license":"ai_generated","mime_type":"image/webp","translations":{"ko":{"alt":"EU 문서를 중심으로 챗봇, 합성 미디어, 얼굴 인식, 일정표와 법적 보호를 연결한 일러스트","caption":"EU AI 규정의 표시 의무와 감독 대상을 시각적으로 정리한 그림이다.","description":null},"en":{"alt":"EU document linked to chatbots, synthetic media, facial recognition, timelines, and legal safeguards","caption":"The illustration summarizes AI transparency duties and oversight under the EU framework.","description":null},"ja":{"alt":"EU文書を中心にチャットボット、合成メディア、顔認識、日程、法的保護を結ぶイラスト","caption":"EUのAI規制における表示義務と監督対象を視覚的に示している。","description":null},"es":{"alt":"Documento de la UE conectado con chatbots, medios sintéticos, reconocimiento facial, plazos y protección legal","caption":"La ilustración resume las obligaciones de transparencia y la supervisión de la IA en la UE.","description":null},"id":{"alt":"Dokumen UE terhubung dengan chatbot, media sintetis, pengenalan wajah, jadwal, dan perlindungan hukum","caption":"Ilustrasi ini merangkum kewajiban transparansi dan pengawasan AI dalam kerangka UE.","description":null},"pt":{"alt":"Documento da UE ligado a chatbots, mídia sintética, reconhecimento facial, prazos e proteção jurídica","caption":"A ilustração resume as obrigações de transparência e a supervisão da IA no quadro da UE.","description":null},"zh-hant":{"alt":"以歐盟文件為中心，連結聊天機器人、合成媒體、人臉辨識、時程與法律保障的插圖","caption":"此圖概述歐盟人工智慧規範下的透明度義務與監管對象。","description":null},"de":{"alt":"EU-Dokument verknüpft mit Chatbots, synthetischen Medien, Gesichtserkennung, Fristen und Rechtsschutz","caption":"Die Illustration fasst Transparenzpflichten und KI-Aufsicht im EU-Rahmen zusammen.","description":null}}}],"published_at":"2026-09-06T22:19:56+09:00","updated_at":"2026-09-06T22:19:56+09:00","license":"cc_by","translation_status":"reviewed","available_locales":["ko","en","ja","es"],"data_locales":["ko","en","ja","es","id","pt","zh-hant","de"],"url":"https://injoys.com/en/articles/eu-ai-act-article-50-transparency-rules-2026"}