At first glance, this sounds perfectly reasonable. After all, who wants to fall for deceptively realistic photos of stylishly furnished, spotless hotel rooms overlooking paradise \u2013 only to arrive at a run-down dump next to a major construction site? Or for apartment listings that leave you wondering, at the viewing, whether you\u2019ve accidentally walked into the wrong building?<\/p>\n\n\n\n
But look closer, and things get complicated. Very complicated.<\/strong><\/p>\n\n\n\n People were manipulating images long before generative AI existed. Throughout history, those who controlled images could shape public opinion \u2013 and, with it, hold power. Stalin famously had purged officials airbrushed out of photographs: the picture changed, and with it the past.<\/p>\n\n\n\n Visual content occupies a special place in human perception. We trust images almost instinctively and rarely question their authenticity \u2013 far more than we trust written text. Or at least, that used to be the case. This is exactly why images have always been one of the most effective tools for deception and manipulation.<\/p>\n\n\n\n What has changed is the scale.<\/strong><\/p>\n\n\n\n With generative AI, deceptively realistic images \u2013 so-called deepfakes<\/em> \u2013 can now be created within seconds, in massive quantities and without any specialized skills. The EU AI Act is the European Union’s response to this new reality.<\/p>\n\n\n\n An estimated 80 million new AI-generated images are created every day worldwide.<\/strong>1<\/a><\/sup><\/p>\n<\/blockquote>\n\n\n\n Unfortunately, these rules also leave many important questions unanswered. This article takes a closer look at these gray areas, contradictions and practical challenges, including questions such as:<\/p>\n\n\n\n Article 50 of Regulation (EU) 2024\/1689 is officially titled as \u00abTransparency obligations for providers and deployers of certain AI systems\u00bb<\/strong>2<\/a><\/sup><\/p>\n\n\n\n Its purpose is to regulate the use of artificial intelligence within the European internal market while protecting safety, fundamental rights, democracy and the rule of law. <\/p>\n\n\n\n One of the most common misunderstandings concerns exactly who these new rules apply to.<\/p>\n\n\n\n Under the AI Act, a provider<\/em> is generally a person or organization that develops \u2013 or has developed \u2013 an AI system and places it on the market under its own name or brand.<\/p>\n\n\n\n Examples include:<\/p>\n\n\n\n According to Article 50(2), these providers must ensure that synthetic image, audio, video and text outputs are marked in a machine-readable format<\/strong> so they can be recognized as artificially generated or manipulated.<\/p>\n\n\n\n This typically refers to technical solutions such as provenance information (a record of where an image came from), embedded metadata, or other robust detection mechanisms.<\/p>\n\n\n\n The AI Act calls this group \u00abdeployers\u00bb \u2013 a term that has caused considerable confusion. The German version of the law uses \u00abBetreiber\u00bb, which sounds like \u00aboperator of an AI service\u00bb. It does not<\/strong> refer to companies operating AI services \u2013 that group has already been covered above.<\/p>\n\n\n\n Instead, it refers to natural or legal persons who use<\/strong> an AI system. In other words, professional users or deployers of AI, such as:<\/p>\n\n\n\n According to Article 50(4), they must disclose when published image, audio, or video content has been generated or manipulated using an AI system.<\/p>\n\n\n\n These obligations do not<\/strong> apply to natural persons who publish AI-generated image, audio, or video content exclusively for personal and non-professional purposes.<\/p>\n\n\n\n For example, a private travel blog would most likely fall under this exemption.<\/p>\n\n\n\n Things become less clear, however, if the blog owner:<\/strong><\/p>\n\n\n\n This is one of many gray areas within the AI Act. \u00abMaking money\u00bb is not necessarily the same as acting in a professional capacity, and the law does not clearly define where that line should be drawn.<\/p>\n\n\n\n Let’s now look at the actual obligation. Under the EU AI Act, both of the groups described above have a duty to disclose AI-generated or AI-manipulated image, video and audio content.<\/p>\n\n\n\n Obvious exceptions include clearly fictional, artistic, or satirical content, as long as it does not<\/strong> claim to truthfully represent real events or real people.<\/p>\n\n\n\n The first group, AI providers, must mark their content in a machine-readable format. The second group, users of AI systems, must label or identify the AI content they publish in a way that people can notice and understand right away, including those using assistive technology. They must also ensure that the machine-readable markings added by the AI generator are not removed, whether intentionally or accidentally.<\/p>\n\n\n\n It can be assumed that publishers are responsible if published content does not meet these requirements. They therefore need to check how the content is actually delivered to users. Image editing processes or the content management system (CMS) used to publish an image can easily remove the AI provider\u2019s machine-readable metadata without anyone noticing.<\/p>\n\n\n\n The EU AI Act includes an exception for AI-based editing that only supports or improves an image and does not substantially alter its semantics.<\/p>\n\n\n\n The key question is therefore: Does the editing only improve the quality of the image, or does it change what the image communicates?<\/p>\n\n\n\n The following AI-based edits would most likely be considered enhancements and would not require disclosure:<\/strong><\/p>\n\n\n\n The following AI-based edits change the meaning or content of an image and would require disclosure:<\/strong><\/p>\n\n\n\n Between these two groups lies a large area of unresolved gray zones.<\/p>\n\n\n\n What happens when AI \u2026<\/strong><\/p>\n\n\n\n We cannot answer these questions precisely, especially not in purely technical terms. There is no fixed number of changed pixels that tells us when the meaning of an image has been substantially altered.<\/p>\n\n\n\n After all:<\/strong><\/p>\n\n\n\n can seriously mislead viewers, depending on the context. By contrast, changing the color of a large part of an image may have almost no effect on its meaning.<\/p>\n\n\n\n Future guidelines, and perhaps soon also courts, will need to clarify where a \u00absubstantial\u00bb change in meaning actually begins.<\/p>\n\n\n\n And there is another question: In a dispute, how can anyone prove with certainty whether an image was generated by AI? Image generators have now become so advanced that even experts can no longer reliably distinguish AI-generated images from real photographs. Take a look at the following images, for example. Which of them are AI-generated? Can you say for sure?<\/p>\n\n\n\n Images 1, 3, and 6 are photographs; images 2, 4, and 5 are AI-generated.3<\/a><\/sup> <\/p>\n<\/details>\n\n\n\n <\/p>\n\n\n\n I consider this one of the regulation\u2019s biggest practical weaknesses. Transparency requirements only work if it is possible to determine whether a piece of content falls under those requirements in the first place.<\/p>\n\n\n\n In many future cases, however, this may be difficult or even impossible if no reliable records of the creation process remain. This issue is likely to become a recurring subject in legal disputes over the coming years.<\/p>\n\n\n\n Article 3(60) of the EU AI Act defines a deepfake as an \u00abAI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful.\u00bb<\/p>\n\n\n\n This definition contains several conditions:<\/strong><\/p>\n\n\n\n That sounds clearer than it actually is, because none of these terms has a precise technical boundary. For example, what exactly does \u201cresembles\u201d mean? Does an image have to depict a specific person or place accurately, or is a general similarity enough?<\/p>\n\n\n\n A synthetic image can be factually accurate even though it is not an authentic record of a real moment. Imagine, for example, that an AI system generates a photorealistic image of Zurich Main Station:<\/p>\n\n\n\n Is this a deepfake? To answer this question, we need to distinguish between factual accuracy and documentary authenticity:<\/p>\n\n\n\n Factual accuracy:<\/strong> The image accurately depicts what a place, person, or object looks like. For the website of the Swiss Federal Railways, this distinction may not matter. The AI generated images shows correctly the Zurich Main Station. <\/p>\n\n\n\n For news reporting, legal evidence, or scientific documentation, however, the distinction is essential. The image does not document an actual moment. <\/p>\n\n\n\n So a synthetic image can be a useful illustration, but it cannot and should not be used as proof that the situation shown actually took place.<\/p>\n\n\n\n Therefore, the disclosure does not mean: This image is false.<\/strong> Instead, it means: This image is not a photographic documentation.<\/strong><\/p>\n\n\n\n That’s because not every synthetic image of a real place or situation is intended to deceive. Therefore, it is questionable whether the negatively charged term ‘deepfake’ is appropriate for a harmless and factually accurate visualisation.<\/p>\n\n\n\n <\/p>\n\n\n\n Another unresolved question concerns entirely AI-generated people.<\/p>\n\n\n\n Imagine that an AI system creates a person without using any real reference image. Since billions of people are alive today or have lived in the past, it is statistically likely that someone somewhere looks very similar to this generated person \u2013 perhaps even several people.<\/p>\n\n\n\n Does that automatically turn the synthetic person into a deepfake?<\/strong><\/p>\n\n\n\n An interpretation based only on resemblance to any random person would be impossible to apply in practice. No company could compare every AI-generated portrait with the entire world population.<\/p>\n\n\n\n A generic synthetic model therefore cannot become a deepfake simply because a very similar-looking person happens to exist somewhere in the world.<\/p>\n\n\n\n Other factors would need to be considered, such as:<\/p>\n\n\n\n This is one of the biggest contradictions in the regulation.<\/p>\n\n\n\n Let’s take a simple example: imagine that a relevant person has been removed from a photograph.<\/p>\n\n\n\n Version A: <\/strong>The person was removed using generative AI. In both cases, the visible result is identical. Yet only the image from Version A must be disclosed as AI-generated or AI-manipulated \u2013 the image from Version B does not. Two identical images are treated differently simply because different tools were used.<\/p>\n\n\n\n The same applies to:<\/strong><\/p>\n\n\n\n In all of these cases, viewers are deliberately misled. But only the AI-generated images must be disclosed.<\/p>\n\n\n\n The EU gives the following reasons for this unequal treatment:<\/p>\n\n\n\n These risks are real. Yet, the new EU rules do not explain why an AI visualization must be disclosed while an identical manual manipulation can remain unlabeled.<\/p>\n\n\n\n A regulation focused on real-world impact should ask more often: Does this image mislead the audience about something that matters? And less often: Which technology was used to create the misleading result?<\/p>\n\n\n\n The EU also provides three optional icons:<\/strong><\/p>\n\n\n\n Using these icons is voluntary. They are not official seals of approval and do not prove legal compliance. Anyone who uses them remains responsible for making sure that the disclosure is sufficient in the specific context.<\/p>\n\n\n\n From a UX perspective, several questions remain:<\/strong><\/p>\n\n\n\n Unfortunately, the EU guidance does not answer these questions in enough detail.<\/p>\n\n\n\n The EU AI Act and its accompanying guidance try to achieve several goals at the same time:<\/p>\n\n\n\n Each of these requirements makes sense on its own. Taken together, however, they are extremely difficult to achieve from a technical perspective, especially on the web.<\/p>\n\n\n\n A label or icon is displayed above or below the image.<\/p>\n\n\n\n Pros<\/strong><\/p>\n\n\n\n Cons<\/strong><\/p>\n\n\n\n The label is permanently burned into the image itself.<\/p>\n\n\n\n Pros<\/strong><\/p>\n\n\n\n Cons<\/strong><\/p>\n\n\n\n The information is stored inside the file in a machine-readable format.<\/p>\n\n\n\n Pros<\/strong><\/p>\n\n\n\n Cons<\/strong><\/p>\n\n\n\n None of these solutions satisfies all of the requirements. In practice, they will probably need to be combined.<\/p>\n\n\n\n It will be interesting to see how the industry approaches this challenge. One thing is already becoming clear: AI disclosure is not just another step during image export. It is evolving into a multi-layered publishing workflow.<\/p>\n\n\n\n For an international media company, that may be acceptable. For a small business, a nonprofit organization, or a single content manager, however, the required workflow seems disproportionately complex \u2013 almost absurd.<\/p>\n\n\n\n Even if all of these requirements are met, we still cannot prevent third parties from:<\/p>\n\n\n\n For this reason, the EU’s goal of permanent labeling cannot realistically be achieved from a technical point of view.<\/p>\n\n\n\n The most realistic objective is simply this: the disclosure should remain intact throughout the export, download and sharing processes that are under the publisher’s control.<\/p>\n\n\n\n But what happens beyond that point?<\/strong><\/p>\n\n\n\n Who is responsible if a perfectly reasonable image compression process removes all metadata? Or if a content management system automatically generates a thumbnail without the required label?<\/p>\n\n\n\n An AI disclosure system does more than simply provide information.<\/p>\n\n\n\n It also creates a mental model. Over time, people may begin to assume that any image without a label must be authentic<\/strong>. That conclusion is both unintended and potentially dangerous.<\/p>\n\n\n\n An unlabeled image may:<\/strong><\/p>\n\n\n\n From a consumer protection perspective, this creates a classic false sense of security<\/strong>.<\/p>\n\n\n\n As a result, unlabeled images may appear more trustworthy than they actually are.<\/p>\n\n\n\n In journalism and political communication, this could have serious consequences. A traditionally manipulated photograph might appear more credible simply because it does not carry an AI disclosure label.<\/p>\n\n\n\n The AI Act is intended to build trust. Ironically, it may also create a new form of misplaced trust.<\/p>\n\n\n\n The EU AI Act applies within the European Union. In most countries outside the EU, there is currently no comparable legal requirement to disclose AI-generated or AI-manipulated images. And simply making a website technically accessible from within the EU is not enough for the AI Act to apply.<\/p>\n\n\n\n However, being based outside the EU does not automatically put a company beyond the reach of these rules. The AI Act is generally relevant for non-EU companies if at least one of the following conditions applies:<\/p>\n\n\n\n At the same time, many countries are working on their own AI regulations. Switzerland, for example, has decided to implement the Council of Europe’s AI Convention: a consultation draft is expected by the end of 2026 and is likely to address issues such as transparency, data protection, non-discrimination and regulatory oversight. Companies outside the EU should therefore also keep an eye on upcoming rules in their own country.<\/p>\n\n\n\n Within the European Union, violations of the AI Act can be enforced by the competent authorities through administrative fines.<\/p>\n\n\n\n For companies outside the EU without offices or legal entities inside the EU, enforcing such fines across borders is likely to be difficult and often not worth the administrative effort. Being located outside the EU does not provide complete legal protection \u2013 but in practice, it seems highly unlikely that a company without any EU presence would actually receive an administrative fine from an EU authority.<\/p>\n\n\n\n Competitors or organizations that are legally entitled to sue may seek to stop the continued use of non-compliant content.<\/p>\n\n\n\n In most cases, companies would first be given an opportunity to correct the issue. Yet, they would still be responsible for the legal costs associated with the complaint.<\/p>\n\n\n\n In some situations, having to stop a marketing campaign or replace published content at short notice may be far more expensive than the fine itself.<\/p>\n\n\n\n Note that even without an EU presence, European competitors may still be able to pursue civil claims before courts in a company’s home country.<\/p>\n\n\n\n Claims for damages are generally possible, but they normally require proof of actual financial harm, causation and the other legal requirements for liability. A mere competitive advantage gained through non-compliance is usually not enough on its own.<\/p>\n\n\n\n The new disclosure requirements do not change copyright law.<\/p>\n\n\n\n The current legal situation for AI-generated content can be summarized as follows:<\/p>\n\n\n\n The main lesson is that we should be cautious before assuming that AI-generated content is free to use. Such content may still be protected by copyright if it has later been integrated into an independent human creative process. It would therefore be risky to assume that an image carrying an AI disclosure label can automatically be used without permission.<\/p>\n\n\n\n Until the EU \u2013 and possibly the courts \u2013 provides clearer answers to the open questions, companies should act carefully and use a transparent decision-making process.<\/p>\n\n\n\n The following six questions should be considered for every image before publication. Only if all six questions are answered with \u00abyes\u00bb would disclosure be required:<\/p>\n\n\n\n The EU AI Act responds to a real problem: AI-generated images can no longer be reliably distinguished from photographs. Highly realistic image manipulation can now be created by almost anyone and distributed with very little effort. In journalism, legal contexts, education and science, transparency about the origin of visual content is clearly useful and can provide real value.<\/p>\n\n\n\n But the current approach in the EU AI Act remains unsatisfactory. <\/strong><\/p>\n\n\n\n The technical implementation of image disclosure on the web has not been properly considered. A label embedded directly in an image remains attached when the file is downloaded, but it is not accessible. An HTML label can be accessible, but it becomes separated from the image when the file is shared.<\/p>\n\n\n\n Small companies, in particular, face the disproportionate challenge of having to manage legal requirements, UX, accessibility, image production and complex technical processes at the same time.<\/p>\n\n\n\n I am especially concerned about the possible side effect of the entire labeling system. Once people expect AI-generated images to carry a label, the absence of a label may be wrongly interpreted as proof that an image is authentic.<\/p>\n\n\n\n The EU AI Act may therefore create new legal uncertainty, major implementation challenges and possibly a dangerous false sense of security.<\/p>\n\n\n\n Useful improvements would include:<\/strong><\/p>\n\n\n\n <\/p>\n\n\n\n Above all, the EU AI Act is a response to a technology that society mainly perceives as a threat. Regulation tends to emerge where uncertainty, loss of control and public pressure come together.<\/p>\n\n\n\n That is precisely why such regulation should be clear, precise and easy to implement.<\/p>\n\n\n\n This article reflects the state of information as of July 2026 and does not constitute legal advice. The European Commission\u2019s final guidelines, <\/a>as well as future regulatory and court practice, may clarify or change individual assessments.<\/p>\n\n\n\n <\/p>\n","protected":false},"excerpt":{"rendered":" A practical guide to Article 50 of the EU AI Act.<\/p>\n","protected":false},"author":1,"featured_media":1622,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"inline_featured_image":false,"footnotes":"[{\"content\":\"https:\/\/imagera.ai\/blog\/ai-image-generation-statistics-2026<\/a>\",\"id\":\"5a76be01-73b6-4552-8870-92e8f06de917\"},{\"content\":\"https:\/\/artificialintelligenceact.eu<\/a>\",\"id\":\"2a6be064-c63c-4f06-8e78-e93d4dd8336f\"},{\"content\":\"https:\/\/www.watson.ch\/quiz\/digital\/138696182-erkennst-du-die-ki-generierten-fake-fotos-hier-kannst-du-dich-testen<\/a>\",\"id\":\"e14188a5-e51e-4e8e-8b3e-6f672df704da\"}]"},"categories":[39,16],"tags":[50,51],"class_list":["post-1620","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence","category-user-experience","tag-eu","tag-regulations"],"_links":{"self":[{"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/posts\/1620","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/comments?post=1620"}],"version-history":[{"count":43,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/posts\/1620\/revisions"}],"predecessor-version":[{"id":1732,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/posts\/1620\/revisions\/1732"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/media\/1622"}],"wp:attachment":[{"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/media?parent=1620"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/categories?post=1620"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/thomas-sokolowski.com\/en\/wp-json\/wp\/v2\/tags?post=1620"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}\n
\n
The legal framework<\/h2>\n\n\n\n
Who does this regulation apply to?<\/h2>\n\n\n\n
Group 1: AI providers<\/h3>\n\n\n\n
\n
Group 2: Professional users of AI<\/h3>\n\n\n\n
\n
Private use is exempt<\/h2>\n\n\n\n
\n
Disclosure requirements<\/h2>\n\n\n\n
Where does image enhancement end and manipulation begin?<\/h2>\n\n\n\n
\n
\n
\n
\n
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/1_echt.webp\")
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/4-unecht.webp\")
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/5-echt.webp\")
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/6-unecht.webp\")
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/8-unecht.webp\")
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/9-echt.webp\")
Click here for see the solution …<\/summary>\n
The deepfake<\/h2>\n\n\n\n
\n
\n
Documentary authenticity:<\/strong> The image is a genuine record of a specific moment or event that actually occurred.<\/p>\n\n\n\nThe accidental lookalike<\/h2>\n\n\n\n
\n
The Photoshop paradox<\/h2>\n\n\n\n
Version B: <\/strong>The person was removed in Photoshop by a human image editor.<\/p>\n\n\n\n\n
\n
The official EU icons<\/h2>\n\n\n\n
\n
![\"\"](\"https:\/\/thomas-sokolowski.com\/en\/wp-content\/uploads\/2026\/07\/ai_labels_row-1024x111.png\")
\n
Always visible, portable and accessible: an almost impossible combination<\/h2>\n\n\n\n
\n
Solution 1: HTML\/CSS labeling<\/h4>\n\n\n\n
\n
\n
Solution 2: Embed the label directly into the image<\/h4>\n\n\n\n
\n
\n
Solution 3: Metadata and Content Credentials<\/h4>\n\n\n\n
\n
\n
Preserved after download \u2013 really?<\/h2>\n\n\n\n
\n
Perhaps the most dangerous side effect<\/h2>\n\n\n\n
\n
What does the EU AI Act mean for countries outside the EU?<\/h2>\n\n\n\n
\n
Legal risks for companies that fail to comply<\/h2>\n\n\n\n
Regulatory fines<\/h5>\n\n\n\n
Injunctions and corrective measures<\/h5>\n\n\n\n
Claims for damages<\/h5>\n\n\n\n
Copyright implications<\/h2>\n\n\n\n
\n
A practical guide<\/h2>\n\n\n\n
\n
My conclusion<\/h2>\n\n\n\n
The rules have several major weaknesses:<\/h4>\n\n\n\n
\n
What should be regulated more clearly?<\/h4>\n\n\n\n
\n
Note<\/h4>\n\n\n\n
Sources and further reading<\/h4>\n\n\n