New EU AI Transparency Law: The Definitive Guide for SMEs
Artificial Intelligence has revolutionized the way small and medium-sized enterprises (SMEs) operate, enabling task automation, improved marketing, and optimized customer service. However, with the great power of AI comes a new legal responsibility. The European Union has just taken a decisive step with the publication of the Code of Practice on Transparency of AI-Generated ContentSynthetic MediaMedia (image, audio, video) artificially created by AI.
This document is an essential practical guide for complying with Article 50 of the AI Act (Artificial Intelligence Act). Its goal is not to hinder innovation, but to ensure that users know when they are interacting with artificial content. If your SME uses AI tools to create or publish content (making you what the law calls a deployer), this article is for you.
The 10 Key Points of the European AI Regulation for SMEs
Section 2 of the document is the most critical for companies that use AI systems. Here we break down the 10 most important points that will change the way you work:
1. Full transparency if you publish AI-generated content
An SME that uses AI to create published images, videos, audio, or text will have to make clear when that content is generated or manipulated by AI, especially if it could be mistaken for real content. This directly affects marketing, social media, advertising, e-commerce, training, corporate communication, and customer service departments. Practical impact: It will no longer be enough to say internally "I used ChatGPT." The legal and ethical problem will arise when the generated content is published and could affect the perception of third parties.
2. Platforms will incorporate automatic marks
AI system providers (such as OpenAI, Midjourney, etc.) must mark generated content in a machine-readable format and make it detectable as artificial. The document specifies the use of metadata, watermarks, and integrated detection systems. Practical impact: SMEs will see that many of their favorite tools start adding marks or labels by default in exported files.
3. Be careful about deleting marks or metadata
The document is emphatic: providers must preserve existing markings and prohibit the intentional removal of marks or metadata by users. Practical impact: If you download an AI-generated image, edit it in Photoshop, and publish it, you must make sure not to remove its provenance metadata. Removing a watermark "because it looks aesthetically unpleasant" can become a direct compliance violation.
4. Visual and auditory content will be the main focus of risk
The regulation pays special attention to synthetic audio, image, and video content. The industry is expected to combine techniques such as cryptographically signed metadata, imperceptible watermarks, and advanced detection systems. Practical impact: If your SME creates ads with fictional models, videos with cloned voices, avatars for e-learning, simulated testimonials, or hyper-realistic product images, you must take extreme transparency measures.
5. Deepfakes require visible and unambiguous labeling
The labeling of deepfakes (ultra-realistic images, audio, or videos generated by AI) must be clear, distinguishable, and available from the very first millisecond a person views the content. Practical impact: If you use a virtual avatar to present your corporate videos, a synthetic voice for your switchboard, or a recreation of a person, the user must know there is an AI behind it without having to search for the information.
6. Official European icons are arriving
The annex of the regulation includes public icons designed by the EU to signal "AI GENERATED" (AI Generated) and "AI MODIFIED" (AI Modified) content. These icons have been tested with real users, demonstrating that those including clear text avoid ambiguities. Practical impact: SMEs will have a standard visual reference. Get used to seeing (and using) these icons on websites, social networks, and advertising campaigns.
7. Hiding labeling is prohibited
The icon or label must be placed perceptibly, be immediately recognizable, be visible for a sufficient amount of time, and, where appropriate, be embedded directly into the content (like a TV channel logo). Practical impact: It will be useless to put a lost note in your website's legal notice or in a hidden dropdown text. Transparency must be upfront and direct.
8. Need for internal processes (even if simple)
Companies are required to have internal processes proportional to their size and resources. This involves documenting how labeling obligations are applied and establishing verification processes. Practical impact: An SME does not need an exclusive legal department for this, but it does need a minimum internal policy: defining when to label, who reviews the content, which tools are permitted, and how an error is corrected.
9. AI literacy for the whole team
The document recommends proportional efforts so that employees and collaborators are aware of transparency obligations. Practical impact: Marketing, sales, communication, design, and customer service departments will have to understand these basic rules. The responsibility no longer falls solely on the "IT technician".
10. Strict editorial control for AI-generated texts
There is a specific commitment regarding human and editorial control of published text. Organizations must have human review policies before publishing. Practical impact: If you publish blog posts, newsletters, product sheets or press releases generated with AI, there must be a physical person responsible for reviewing and approving that publication (Human-in-the-loop).
Recommended Workflow for SMEs
To visualize what the publishing process in your company should look like from now on, review the following diagram:
Action Plan: What your SME should do today
The positive reading of this regulation is clear: the EU is not going to prohibit SMEs from using AI. On the contrary, it seeks to facilitate trustworthy use. Using AI will be normal; hiding it when it affects the public will be the real problem.
We recommend creating today a 1-page internal policy that includes:
- Inventory: Which AI-created content is regularly published.
- Labeling rules: When and how it should be labeled (using the EU icons).
- Approved tools: Which AI software is allowed to be used in the company.
- Responsible parties: Who reviews and approves the content before publishing it.
- Red lines: What cannot be done (deleting watermarks, simulating real testimonials with AI, publishing deepfakes without notice).
- Correction protocol: How to act if unlabeled content is published by mistake.
Frequently Asked Questions (Q&A)
Does this law prevent me from using ChatGPT to write internal emails?
No. The law focuses on the transparency of content that is published or that interacts with third parties (customers, users, general public). Internal use to summarize texts or brainstorm strategies does not require public labeling.
What happens if I hire an external agency for my marketing and they use AI?
As the company that publishes the content on its channels (the deployer), your SME remains responsible before the law.
You must require your providers and agencies to comply with transparency regulations and inform you if they deliver AI-generated material.
Do I have to label a stock photo that was generated with AI?
Yes. If you acquire a synthetic image from an image bank and use it in an advertising campaign, you must maintain the metadata and, depending on its realism and context, add the corresponding visual label to avoid misleading the consumer.
Turn Transparency into your Competitive Advantage
Adapting to the new EU AI Act doesn't have to be a legal headache. In fact, transparency builds trust, and consumers increasingly value honest brands.
At Netretina and through our NAiOS system, we see this regulation as a clear opportunity for SMEs. We help you integrate Artificial Intelligence into your business processes in an ethical, legal, secure, and highly profitable way. Don't let legal fear hold back your innovation.
Do you want to implement AI in your SME while complying with all European regulations without complications? Contact our experts today by writing to info@netretina.ai and discover how we can drive your business into the future.
