Transparency obligations under Article 50 of the EU AI Act: What industrial and MedTech companies must comply with from August 2, 2026

As of August 2, 2026, providers and deployers of certain AI systems must comply with the transparency obligations set out in Article 50 of the EU AI Act (Regulation (EU) 2024/1689). For industrial companies and MedTech manufacturers, this is the first direct point of contact with the AI Act, regardless of whether their actual AI applications are already subject to stricter rules.

"From now on, everything created with AI must be labeled." You have been reading this sentence for weeks. It is incorrect. Article 50 does not contain a blanket labeling obligation, but rather four distinct cases with different addressees and several exceptions. This article shows what this specifically means for an industrial company and a MedTech manufacturer, with examples from both sectors.

What does Article 50 of the EU AI Act require?

Article 50 regulates four transparency obligations: notification of AI interaction, labeling of AI-generated content, disclosure regarding emotion recognition, and labeling of deepfakes and certain AI-generated texts.

The four cases, each with the responsible party:

1. Notification of AI interaction (Para. 1: Provider): AI systems that interact directly with humans must be designed so that the person realizes they are communicating with an AI. Exception: when this is obvious from the context.
2. Labeling of synthetic content (Para. 2: Provider): Providers of generative AI must mark generated image, audio, video, and text content in a machine-readable format as artificially generated or manipulated.
3. Disclosure regarding emotion recognition (Para. 3: Deployer): Anyone using a system for emotion recognition or biometric categorization must inform the affected individuals.
4. Labeling of deepfakes and AI texts of public interest (Para. 4: Deployer): Deployers must disclose artificially generated or manipulated image, audio, and video content, as well as AI-generated texts concerning matters of public interest.

Provider or deployer: who is subject to which obligation?

Providers develop or make AI systems available and bear the obligations under paragraphs 1 and 2; deployers use them under their own name and bear the obligations under paragraphs 3 and 4.

Both industrial companies and MedTech manufacturers are in most cases deployers: they rarely develop their own generative models, but they do use purchased AI systems, chatbots, text and image generators, and sometimes emotion recognition. For them, the deployer obligations are the priority, while the technical labeling according to paragraph 2 lies with the provider of the respective tool. It is therefore important to coordinate with the providers used: does their system already label generated content in a machine-readable way, or must you ensure the disclosure yourself?

What does Article 50 mean for industrial companies?

For industrial companies, Article 50 primarily concerns customer-facing and public-facing AI: chatbots in technical support, AI-generated product and marketing texts, and AI images.

Concrete examples from everyday industrial life:

• Chatbots in technical customer service: Falls under paragraph 1. Users must be aware that they are communicating with an AI and not a service technician.
• AI-generated product descriptions, datasheet texts, or blog posts: If such texts are published on matters of public interest, paragraph 4 applies, unless they undergo a genuine editorial review with clear assignment of responsibility.
• AI images in marketing and sales: As synthetic content, they fall under the labeling and disclosure logic of paragraphs 2 and 4.

A purely internal AI knowledge system. For example, for researching technical documents, without direct contact with individuals and without published content – generally does not trigger any labeling obligation under Article 50.

What does Article 50 mean for MedTech companies?

For MedTech manufacturers, Article 50 concerns customer- and patient-facing AI, not the clinical AI within the medical device itself, which is regulated separately as high-risk AI.

This distinction is crucial because it is often blurred in practice:

• Patient- or customer-facing chatbots, such as those on a website or support portal, fall under paragraph 1: A notice regarding the AI interaction is mandatory.
• AI-generated content in communication, training, or education falls under paragraphs 2 and 4, depending on the context, with the editorial exception for verified texts.
• The AI function within the medical device itself, such as a diagnostic algorithm, is not a matter for Article 50. It is subject to the high-risk regime of the AI Act and its interaction with the MDR or IVDR, for which market surveillance in Germany lies with the BfArM. This is a separate issue and not the subject of transparency obligations.

For MedTech manufacturers, this means: As of August 2, 2026, transparency obligations are the part that must be fulfilled immediately and independently of product approval, wherever AI interacts directly with people or generates published content.

Do I really have to label all AI content?

No. Article 50 provides for several exceptions, for art and fiction, for editorially verified content, and for purely supportive or insignificant edits.

The most important exemptions for both industries:

• Editorial review: AI-generated content on matters of public interest is exempt if it undergoes human review with clear accountability. A simple "quick read" is not sufficient.
• Insignificant editing: Systems that process input only in a supporting capacity or without significant modification are exempt.
• Art, satire, fiction: Disclosure must not interfere with the presentation of the work.
• No retroactive effect: There is no retroactive labeling requirement for deepfakes created before August 2, 2026.

Assess on a case-by-case basis rather than labeling everything as a precaution. Unnecessary notices do not create security; they only dilute the impact where labeling truly matters.

Deadlines, fines, and jurisdiction

The transparency obligations have been in effect since August 2, 2026, and are enforceable from that date; a transition period until December 2, 2026, applies only to the technical marking of generative systems already on the market.

The active enforcement phase begins on August 2, 2026, and the competent authorities will commence market surveillance. Violations of Article 50 can be penalized with fines of up to 15 million euros or 3% of total worldwide annual turnover, whichever is higher; for SMEs, the lower amount applies. In Germany, the Federal Network Agency (Bundesnetzagentur) assumes central oversight and operates an AI service desk; for AI-based medical devices, market surveillance remains with the BfArM.

What should industrial and MedTech companies do now?

Inventory the AI systems in use, identify the relevant Article 50 cases, and establish clear notices, responsibilities, and documentation.

Four steps:

1. Create an AI inventory. Document which AI applications are in use and which of those generate content, interact with humans, or recognize emotions.
2. Clarify your role. Are you the provider or the operator of the system in question? MedTech manufacturers should clearly distinguish between customer/patient-facing AI (Article 50) and clinical AI within the product (high-risk regime).
3. Implement notices and processes. Create clear AI guidelines for chatbots and published content, define responsibilities, and update workflows as well as contracts with agencies or freelancers.
4. Keep evidence auditable. Document which content is labeled, how it is labeled, and where the editorial review takes place.

This is exactly where MAIA comes in. Whether it is technical documentation in mechanical engineering or regulatory documents in the MedTech sector: verified answers and a seamless audit trail make disclosure and labeling verifiable. MAIA provides answers that can be traced back to the specific document version and page; customer data is not used for training, and the platform is designed for European data protection and compliance requirements, Made in Germany & Switzerland. For an industrial or MedTech company that needs to document its AI usage and keep it auditable, this is the foundation for practically fulfilling the obligations under Article 50.

FAQ

Since when have the transparency obligations under Article 50 of the EU AI Act been in effect? Since August 2, 2026. For machine-readable labeling of generative systems that were already on the market before this date, there is a transition period until December 2, 2026.

What does Article 50 mean for industrial companies? It primarily affects chatbots in technical support, AI-generated product and marketing texts, and AI images. Purely internal AI knowledge systems without customer contact and without published content generally do not trigger a labeling requirement.

What does Article 50 mean for MedTech companies? It affects customer- and patient-facing AI such as chatbots and published AI content. Clinical AI within the medical device itself does not fall under Article 50, but rather under the high-risk regime of the AI Act in conjunction with the MDR or IVDR (market surveillance by the BfArM).

Do I have to label all AI content? No. Article 50 regulates four specific cases. There are exceptions for art, satire, fiction, and editorially reviewed content.

What fines are imposed for violations of Article 50? Up to 15 million euros or 3% of the total worldwide annual turnover, whichever is higher. For SMEs, the lower of the two values applies.

About MAIA

MAIA is the AI knowledge platform for industry and MedTech in the DACH region. It unlocks technical and regulatory documents, checks requirements against internal records, and provides answers that are traceable down to the document version and page – exactly the level of traceability required in practice by Article 50 and the underlying documentation obligations. Customer data is not used for training, and the platform is designed for European data protection and compliance requirements, Made in Germany & Switzerland. Would you like to see how you can document your AI usage and keep it auditable? Book a free demo.

‍

Sources

Federal Network Agency. (2026). AI Service Desk for the implementation of the EU AI Act. Bonn: Federal Network Agency.

European Union. (2024). Regulation (EU) 2024/1689 (AI Act), Article 50. Official Journal of the European Union.

Fraunhofer Academy. (2026, July). Transparency obligations under Art. 50 of the AI Act. https://blog.academy.fraunhofer.de

IHK Schleswig-Holstein. (2026, July). New transparency obligations starting August 2026 – what companies should check now.https://www.ihk.de/schleswig-holstein

TÜV Rheinland Consulting. (2026). Transparency obligations in the EU AI Act – Article 50. https://consulting.tuv.com

usd AG. (2026, July). AI Omnibus: The new deadlines of the EU AI Act. https://www.usd.de

‍