We use privacy-by-default, cookieless analytics.

Optional analytics cookies are used only if you accept them.

Read our Terms of Use and Privacy Policy.

Article 50 GuideEU AI Act

EU AI Act Article 50 When must you disclose AI?

Tell people when they interact with AI, encounter specified biometric uses or see content covered by .

Providers handle interaction notices and synthetic-output marking. Deployers handle notices and disclosures for their own covered uses.

Back to main artifactReview sources
Author
Sorena AI
Published
May 9, 2026
Updated
Oct 21, 2026
6 sections7 primary sources
Overview

assigns different duties to the company providing a system and the organisation using it. Check the function, audience and exception before choosing a notice or label. These duties can apply alongside high-risk AI requirements.

Section 1

1. Find your disclosure duty

Providers of systems intended to interact directly with people must ensure people know they are interacting with AI, unless an exception applies. Providers of systems generating synthetic content must also handle machine-readable marking and detection.

Deployers have separate duties for emotion and biometric systems, deepfakes and specified AI text published to inform the public. Provide clear, distinguishable and accessible information by the first interaction or exposure.

has applied since 2 August 2026. Only Article 50(2) marking and detection for systems placed on the market before that date has a transition to 2 December 2026.

  • Interaction notice: tell people they are interacting with AI unless it is obvious to a reasonably well-informed, observant and circumspect person in context or a stated exception applies.
  • Synthetic-output marking: providers mark generated audio, images, video and text in a machine-readable and detectable form. Technical solutions must meet the feasibility conditions.
  • Biometric notice: deployers inform people exposed to an or .
  • Content disclosure: deployers disclose images, audio or video and covered public-interest text, subject to exceptions.
  • Timing and access: give information by the first interaction or exposure and meet applicable accessibility requirements.
Section 2

2. Tell people when they are interacting with AI

For a chatbot, voice assistant, agent or support flow, show the notice by the first interaction. Check any claimed obviousness or law-enforcement exception against (1).

To rely on obviousness, record what the relevant user would understand from the interface and context. Consider the actual user group.

The law-enforcement exception requires legal authorisation to detect, prevent, investigate or prosecute crime and appropriate safeguards for third-party rights. It does not apply to systems available to the public for reporting a criminal offence.

Operational example: "You are chatting with an AI assistant" at the chat entry point. Save the deployed wording, placement, language and accessibility checks to show that users received it in time.

  • Place the notice at the chat entry, voice prompt, intake form or other first interaction.
  • Use direct wording without overstating capability, accuracy, autonomy or human review.
  • Check every supported language, channel and assistive technology.
  • Save screenshots, release notes, copy approvals and accessibility checks.
  • For high-risk systems, also prepare the separate Article 13 instructions supplied to deployers.
Section 3

3. Mark outputs and disclose covered content

Providers must ensure synthetic audio, image, video and text outputs are machine-readable and detectable as AI-generated or manipulated. Technical solutions must be effective, work across systems, remain reliable and resist failure as far as technically feasible, accounting for content type, costs and the state of the art.

(2) exempts standard assistive editing or changes that do not substantially alter input data or meaning. It also exempts legally authorised use for detecting, preventing, investigating or prosecuting crimes. Save the function, legal basis and output comparisons for any exception.

Deployers must disclose images, audio and video. For evidently artistic, creative, satirical, fictional or similar works, disclose the existence of AI generation or manipulation in an appropriate way that preserves the display or enjoyment of the work.

Deployers must also disclose AI-generated or manipulated text published to inform the public on matters of public interest. Check the law-enforcement exception or the exception for human review/editorial control with a person holding editorial responsibility.

The Commission's optional EU icons are free to use in SVG or PNG without attribution. Choose the appropriate icon after checking the legal trigger and exception. An icon alone does not establish compliance or Code membership.

  • Provider evidence: marking design, detectability tests, content-type limits, interoperability assumptions and release controls.
  • evidence: content asset, audience, channel, disclosure words, placement, timing and any creative-work analysis.
  • Public-interest text: save its publication purpose, human review or editorial control, responsible person and approval supporting any exception.
  • Keep provider technical marking and deployer public disclosure checks separately; both can apply.
  • For the June 2026 Code, retain signature and implemented commitments. Its adequacy assessment covers Articles 50(2), (4) and (5); adherence is not conclusive proof of compliance.
  • Choose Fully AI-Generated for an entire covered asset apart from prompting; Partially AI-Modified for human-made content altered into a or covered text; or the basic icon with a custom label or explanation. Black, white and 50%-transparent versions are available.
  • Check the first exposure: readable size, plain words, no intervening overlays, sufficient display time, and alt text or an ARIA label where possible. Make any second information layer accessible.
  • Code signatories must meet Section 2 placement commitments. The Commission summary calls for embedding the icon except for creative works, unless an equivalent alternative such as an overlay is available, and retaining visibility after sharing/download. Check the export and reshared copy.
Section 4

4. Notify people exposed to biometric uses

Deployers must inform people exposed to an or . Check that the use is lawful before deployment; a notice does not authorise a prohibited use.

Personal data remains subject to the applicable EU data-protection rules. Check the lawful basis, data categories, safeguards and documentation alongside the notice.

  • Identify exposure: camera zone, kiosk, workplace, education setting, access control or customer interaction.
  • Provide consistent AI and privacy notices by exposure; check the distinct requirements for each.
  • (3) exempts legally permitted crime detection, prevention or investigation with appropriate third-party-rights safeguards and compliance with EU law. Keep the legal basis and safeguards; other Article 50 exceptions have different wording.
  • Save signage, interface notices, privacy or data protection impact assessment (DPIA) references, supplier specifications and deployment maps.
  • Complete Article 5 prohibited-use and Annex III high-risk checks as well.
Section 5

5. Keep high-risk instructions alongside notices

A high-risk system can need notices and Article 13 instructions for deployers. Article 50(6) preserves the high-risk obligations.

Article 13 instructions must be concise, complete, correct, clear, relevant, accessible and understandable. They help deployers interpret outputs and use the system appropriately.

Article 26 requires deployers to follow instructions, assign competent oversight, monitor operation and retain logs under their control. Keep evidence of how the instructions translate into operating procedures.

  • : user and audience notices, labels and required technical marking.
  • Article 13: purpose, capabilities, limits, output interpretation, oversight, maintenance, logging and relevant risks in provider instructions.
  • Article 26: operating procedures, competent oversight, monitoring, escalation and controlled-log retention.
  • Link the records when several duties apply to the same system.
  • Review roles and instructions after a substantial modification or changed intended purpose.
Section 6

6. Save evidence and check guidance

Before release or publication, confirm who must receive which notice, when they receive it and how you can prove delivery.

The final voluntary Transparency Code was published on 10 June 2026. The Commission found it adequate for Articles 50(2), (4) and (5) on 8 July; the AI Board adopted its assessment on 9 July. Adherence does not conclusively establish compliance.

The Commission published final non-binding guidelines on 20 July 2026. Use them to interpret the duties while keeping the Regulation as the controlling law. Article 50(6) also preserves other EU and national transparency duties.

Does a chatbot notice replace high-risk instructions?

No. covers specified notices and marking. A high-risk system also needs Article 13 deployer instructions and Article 26 operating controls.

Who discloses a ?

The deployer using the system to generate or manipulate the content must disclose that it is artificial, subject to the specific exceptions and special treatment for creative and similar works.

What marking evidence should a provider keep?

Keep the design, machine-readable format, detectability tests, content-type limits, reliability and interoperability assessment, release approval and any exception analysis.

What must exposed people be told about biometric systems?

Inform them that the emotion recognition or is operating by the time of exposure, unless the (3) exception applies. Meet separate personal-data duties too.

Do all duties have a grace period?

No. Only (2) marking and detection for systems placed on the market before 2 August 2026 has a transition to 2 December 2026. The other Article 50 duties apply from 2 August 2026.

  • Identify the function: interaction, content generation, biometric exposure, or covered public-interest publication.
  • Assign the provider or deployer and the audience, timing and delivery channel.
  • Approve notice words, placement, languages, accessibility and any exception.
  • Save screenshots, marking specifications and tests, publication records, notices, editorial approvals, privacy assessments, instructions and operating procedures.
  • Review after new content types, channels, users, models, risk classification, revised guidance or changes to the Code's adequacy status.
Get help

Review your AI notices and labels

Sorena can help check your disclosure duties, exception analysis and release evidence.

Research workflow
Ask Research Copilot

Ask about an interaction notice, output marking, biometric use or publication disclosure.

Explore next step
Talk to Sorena
Discuss your disclosure review

Review your wording, placement, evidence and high-risk instructions with Sorena.

Explore next step
Primary sources

References and citations

Regulation (EU) 2024/1689 (Artificial Intelligence Act)
eur-lex.europa.eu7 citations
Referenced sections
  • Primary legal text used for Article 50 transparency duties, Article 13 high-risk instructions for use, Article 25 role changes, and Article 26 deployer obligations.
  • Article 50 is the legal source for provider interaction notices, synthetic-content marking, deployer notices, deepfake disclosure, public-interest text disclosure, timing, accessibility, and exceptions.
  • Article 50(1) supports the direct-interaction notice rule and Article 50(5) supports timing, clarity, distinguishability, and accessibility expectations.
Show 4 more
  • Article 50(2) supports provider machine-readable synthetic-content marking; Article 50(4) supports deployer disclosure duties for deepfakes and certain public-interest text.
  • Article 50(3) supports deployer notice duties for exposed natural persons and the link to applicable EU personal-data rules.
  • Articles 13, 25, 26, and 50 support the distinction between user-facing transparency duties, deployer instructions for high-risk AI, operating duties, and role changes along the AI value chain.
  • Articles 13, 25, 26, and 50 support the checklist fields for triggers, actor allocation, timing, high-risk instructions, deployer operation, exceptions, and evidence.
European Commission - AI Act regulatory framework
digital-strategy.ec.europa.eu4 citations
Referenced sections
  • Commission overview used for public explanation of transparency risk, chatbot awareness, AI-generated content identification, deepfake labeling, and the AI Act risk framework.
  • Commission overview describing transparency risk, chatbot awareness, identifiable AI-generated content, deepfake labeling, and the AI Act risk-based framework.
  • Commission overview gives chatbot interaction as an example of transparency risk where people should know they are interacting with a machine.
Show 1 more
  • Commission overview identifies emotion recognition and biometric categorisation in the AI Act risk framework, including prohibited and high-risk examples.
European Commission - Code of practice on marking and labelling AI-generated content
digital-strategy.ec.europa.eu3 citations
Referenced sections
  • Commission page for the final June 2026 voluntary transparency code, its marking and labelling measures, sign-up route, and adequacy-assessment status.
  • Commission page records the final June 2026 voluntary code, provider marking and deployer labelling commitments, sign-up route, and EU icons.
  • Commission page supports the final voluntary code's provider and deployer measures, sign-up context, EU icons, and adequacy-assessment status.
European Commission - Article 50 transparency questions and answers
digital-strategy.ec.europa.eu2 citations
Referenced sections
  • Official final-guidance companion explaining scope, examples, exceptions, evidence routes, and the limited Article 50(2) grace period.
  • Commission companion questions and answers on Article 50; the legal duties stated here are grounded in Articles 3 and 50 and the 2026 amendment.
European Commission - EU icons for labelling AI-generated content
digital-strategy.ec.europa.eu2 citations
Referenced sections
  • Official icon downloads, selection, licence, scope, accessible placement and Code commitments; updated 13 October 2026.
  • Explains icon selection, SVG and PNG downloads, free use without attribution, optional status, Article 50(4) scope and exceptions, accessibility, and the Code's placement commitments.
European Commission - Final Article 50 transparency guidelines
digital-strategy.ec.europa.eu2 citations
Referenced sections
  • Final non-binding Commission guidelines published on 20 July 2026 for interpreting the scope and implementation of Article 50 duties.
  • Final non-binding Commission guidance published on 20 July 2026 supports the scope, actor, exception, timing, and implementation distinctions used in this section.
European Commission opinion on the Article 50 Code of Practice
digital-strategy.ec.europa.eu2 citations
Referenced sections
  • Official source for the Commission and AI Board adequacy assessments and the statement that adherence is not conclusive evidence of compliance.
  • Records the Commission's 8 July 2026 adequacy conclusion, the AI Board's following-day assessment, and the warning that adherence is not conclusive evidence of compliance.

Related guides

Common questions, clear answers | EU AI Act

Clear answers on AI Act scope, risk categories, provider and user duties, general-purpose models, disclosures, rights, assessments and deadlines.

EU AI Act deadlines and compliance calendar

Find the AI Act deadline for your system or model. Check high-risk dates, transparency duties, older-model transitions and the evidence to prepare.

EU AI Act fundamental rights impact assessments

Check who needs an Article 27 fundamental rights impact assessment, what it must contain and how DPIA reuse, notification and registration work.

EU AI Act GPAI model evidence checklist

Collect model documentation, integration information, copyright records and training summaries. Add testing, incident and security records for systemic risk.

EU AI Act monitoring and serious incidents

Set up high-risk AI monitoring, identify serious incidents and check reporting deadlines, recipients, deployer escalation and corrective action.

EU AI Act penalties and fines

Check the AI Act fine tiers, smaller-organisation caps, GPAI provider fines and the facts authorities consider when deciding a penalty.

Show all related guides (34)

EU AI Act serious incident reporting steps

Check serious-incident harm, preserve evidence and choose the reporting deadline and recipient for high-risk AI systems and systemic-risk GPAI models.

EU AI Act timeline and phased deadlines

See which AI Act rules apply now and what comes next. Check high-risk dates, older-model transitions and the difference between legal and standards dates.

EU AI Act vs ISO/IEC 42001

Compare AI Act legal duties with ISO/IEC 42001 management-system requirements. See which records can be reused and what certification does and does not cover.

EU AI Act vs NIST AI RMF

Compare AI Act legal duties with the voluntary NIST AI RMF. Map risk records to specific requirements and identify legal steps that need separate evidence.

EU AI Act: Check whether the Act applies

Use five checks to establish AI Act scope, exclusions, your role, risk categories and evidence for the decision.

EU AI Act: Choose a high-risk assessment route

Choose between provider-run internal control, notified-body assessment and a product-law procedure before release.

EU AI Act: Classify a tool before approval

Classify an AI tool or model before purchase or release. Check EU scope, prohibited uses, high-risk rules, transparency and GPAI duties.

EU AI Act: Compliance checklist

Check each AI tool, decide which duties apply, and give every action an owner and evidence link.

EU AI Act: Conformity assessment

Some high-risk systems use a provider-run assessment. Others need an authorised independent assessor or a product-law assessment.

EU AI Act: Does it apply to you?

Check whether the AI Act applies to your system or model, identify your provider or deployer role, and keep evidence for the decision.

EU AI Act: Duties for general-purpose AI models

Check GPAI provider duties for documentation, integrator information, copyright and training summaries, plus extra requirements for systemic-risk models.

EU AI Act: High-risk requirements

High-risk providers must meet system requirements before supply. Deployers must use and monitor the system under the provider's instructions.

EU AI Act: High-risk system checklist

Check the system's risk controls, data, documentation, logs, instructions, human oversight and testing before release.

EU AI Act: Is your use case high-risk?

Find high-risk AI examples in recruitment, banking, education, healthcare, public services and products. Check Article 6 conditions and exceptions.

EU AI Act: Plan your compliance work

Build a compliance plan for each AI tool: your role, the duties that apply, the evidence needed and the person responsible.

EU AI Act: Standards and quality management

Use EN 18286:2026 to organise provider quality-management procedures and keep evidence that they work for each high-risk system.

EU AI Act: Technical file and evidence templates

Build the technical file and provider records for a high-risk AI system, with owners and evidence links for each field.

EU AI Act: Which AI uses are banned?

Check Article 5 prohibited AI practices, the narrow exceptions, evidence to retain and the new content prohibitions from 2 December 2026.

Are you the provider or the deployer? | EU AI Act

Check provider, deployer, importer and distributor roles, when changes create provider duties, and how model and system responsibilities differ.

Do you need a notified body? | EU AI Act

Find the assessment route for high-risk AI, when a notified body is required, and which declaration, CE-marking and registration records you need.

Does this tool count as AI? | EU AI Act

Check whether software counts as AI, how models differ from systems, when embedded AI is high-risk, and which scope and role evidence to keep.

How do you monitor AI after release? | EU AI Act

Understand high-risk AI monitoring under Article 72: the provider plan, deployer feedback, logs, corrective action, incidents and change reviews.

Is this industry use high-risk? | EU AI Act

See which industry AI uses fall under Annex III, when Article 6(3) can apply, and what providers must document and register.

What belongs in the technical file? | EU AI Act

See what Article 11 and Annex IV require in high-risk AI technical documentation, from system identity and testing to monitoring and retention.

What must general-purpose model providers do? | EU AI Act

Check general-purpose model duties, systemic-risk classification, open-source conditions, copyright, training summaries, incidents and application dates.

When must you disclose AI use? | EU AI Act

Check AI interaction notices, technical content marking, biometric notices, deepfake and text disclosures, exceptions, accessibility and deadlines.

When must you report a serious incident? | EU AI Act

Check serious-incident outcomes, reporting roles, 15-, 10- and two-day deadlines, authority recipients, corrective action and the separate model route.

Who needs a rights impact assessment? | EU AI Act

Check who needs an Article 27 fundamental rights impact assessment, what it must cover, and how privacy assessments, notification and registration fit.