Public Applicability Quick Check

QCK-EUAI-ART50-001 | ODA3 Institute | Public Quick Check | Version 1.0 | Published | 5 August 2026

Purpose

Use this Quick Check to identify whether an AI system, product, feature or content workflow may engage one or more EU AI Act Article 50 transparency obligations. Complete it separately for each system or materially distinct use case.

This is an initial applicability screen. It does not determine compliance, replace a system-specific legal analysis or test whether a control works in practice.

Before you begin

Record:

• System, product or workflow name
• Business owner and technical owner
• Jurisdictions and user populations served
• Whether your organization provides the system, deploys it, or may do both. An organization may act as provider, deployer or both, and the applicable role may differ by system, component, use case and Article 50 obligation.
• Principal third-party models, platforms and content tools
• Date first placed on the market, if relevant

Answer each question Yes, No or Unsure. Treat Unsure as an action item—not as a negative answer.

The eight-question check

1. EU connection

Is the system placed on the EU market, put into service in the EU, used by an organization in the EU, or producing outputs used in the EU?

• Yes / Unsure: Continue through every question.
• No: Record the territorial basis for that conclusion. Obtain legal review where EU reach remains plausible.

2. Provider, deployer or both

Has the organization documented — with supporting rationale — whether it acts as the provider, the deployer, or both for this specific system and use case, and allocated each disclosure, marking, evidence and response responsibility accordingly?

• No / Unsure: Responsibility is unresolved. Third-party procurement does not by itself settle the organization's role or remove deployer obligations. Role can also differ per obligation within the same system — resolve this before answering Questions 3–7.
• Yes: Confirm the allocation in contracts, operating procedures and evidence ownership.

3. Direct interaction with people

Is the AI system intended to interact directly with a person—for example through a chatbot, voice interface, virtual assistant, agent or conversational feature?

• Yes / Unsure: Assess whether the person is clearly informed that they are interacting with AI.
• Do not assume the disclosure exception applies merely because AI is common or the interface appears automated. The Commission directs that the "obviousness" exception be interpreted restrictively, from the standpoint of a reasonably well-informed, observant and circumspect person — not the deploying organization's own view of what is obvious.

4. Synthetic content generation or manipulation

Does the system generate or manipulate audio, image, video or text content?

• Yes / Unsure: Assess the provider-side machine-readable marking and detectability obligation, including technical feasibility and the state of the art.
• Identify whether the activity is only assistive editing or does not substantially alter the input or its semantics; do not apply that boundary without documenting the facts. Assess assistive-editing boundaries separately for each applicable Article 50 obligation.

5. Deepfakes

Does your organization generate, publish or otherwise deploy AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and could falsely appear authentic or truthful to a reasonably well-informed person?

• Yes / Unsure: Assess the deployer-side deepfake disclosure requirement and any applicable artistic, creative, satirical, fictional or analogous-work treatment. This treatment reduces how the disclosure must be presented (it need not hamper the display or enjoyment of the work); it does not remove the disclosure obligation entirely.
• This is primarily a deployer obligation, triggered at time of publication or deployment. Confirm against your Question 2 role determination before concluding the obligation does not apply.

6. Public-interest text

Does your organization publish AI-generated or manipulated text intended to inform the public on matters of public interest?

• Yes / Unsure: Assess the deployer-side disclosure requirement.
• Record whether the content underwent human review or editorial control and whether a natural or legal person holds editorial responsibility and takes responsibility for publication. Both elements must be evidenced, not just claimed.

7. Emotion recognition or biometric categorisation

Does the system perform emotion recognition or biometric categorisation on a person?

• Yes / Unsure: The obligation rests with the deployer to make the exposed person aware of the system's operation, and to address applicable data-protection requirements.
• Separately screen for prohibited-practice and other AI Act issues; this Quick Check covers Article 50 transparency only.

8. Pre-2 August 2026 transition claim

Is the organization relying on the limited transition to 2 December 2026 for Article 50(2) marking and detection duties for a qualifying system placed on the market before 2 August 2026?

• Yes / Unsure: Document the system, placement date, applicable provision, legal basis and evidence supporting the transition claim.
• This limited transition derives from the AI Omnibus grandfathering amendment and applies only to qualifying Article 50(2) marking and detection duties. It does not postpone obligations under Article 50(1), 50(3) or 50(4), all of which remain live from 2 August 2026 regardless of when the system was placed on the market.
• The transition is not a general Article 50 grace period.

Read your result

Potential Article 50 applicability identified

Use this result if any of Questions 3–7 is answered Yes or Unsure. Complete a system-specific provider/deployer analysis, identify the applicable obligation and exception boundaries, collect implementation evidence and test whether the relevant control works through the real user or content lifecycle.

Responsibility or transition analysis incomplete

Use this result if Question 2 or 8 is No or Unsure. Resolve ownership, legal role, third-party dependencies and any transition claim before management relies on the readiness position.

No trigger identified in this initial screen

Use this result only where Questions 3–7 are all No and the EU-connection analysis is documented. This is not a finding that Article 50 is inapplicable. This conclusion is valid only for the current version, use case, and market. Re-screen when the system, use case, content type, market, model, vendor or deployment changes.

Minimum next-step record

For every system screened, retain:

1. Completed answers and rationale
2. Provider/deployer determination
3. Applicable trigger, exception and transition analysis
4. Disclosure or marking-control owner
5. Evidence repository/location and last verification date
6. Known limitations and unresolved questions
7. Required legal, technical or governance follow-up

Notably Absent

This Quick Check does not contain scoring, a compliance threshold, technical test procedures, a finding-severity method or certification criteria. It does not establish that disclosures are sufficiently clear, that machine-readable marks remain detectable, that an exception applies, or that evidence is adequate for a regulator. Those questions require system-specific analysis and testing.

Required limitation

Organization-led self-assessment using the ODA3 Article 50 Transparency Evidence Readiness Methodology. It does not constitute independent assurance, certification, regulatory approval, legal advice or a determination of compliance.

Methodology Note

This Quick Check is derived from ODA3 Institute's closed Stage 0 evidence baseline and public European Commission materials, including the final Article 50 Guidelines adopted 20 July 2026, the Commission's Article 50 Q&A and the dedicated AI Act enforcement framework. ODA3 Institute holds no proprietary enforcement, complaint or non-compliance dataset for this subject.

Copyright © 2026 ODA3 Pvt Ltd. All rights reserved.