{"id":1084,"date":"2026-09-05T20:32:38","date_gmt":"2026-09-05T18:32:38","guid":{"rendered":"https:\/\/beeclear.pl\/en\/?p=1084"},"modified":"2026-09-27T14:59:27","modified_gmt":"2026-09-27T12:59:27","slug":"labelling-ai-content-and-the-ai-act","status":"publish","type":"post","link":"https:\/\/beeclear.pl\/en\/labelling-ai-content-and-the-ai-act\/","title":{"rendered":"Labeling AI content in practice: what the AI Act actually requires from graphic designers and video creators"},"content":{"rendered":"\n<p class=\"is-style-lede wp-block-paragraph\">A lot of fear has built up around the EU&#8217;s AI Act: \u201cnow every AI image needs a visible label.\u201d In practice, the law is much narrower \u2013 a visible \u201cGenerated by AI\u201d marker applies almost exclusively to photorealistic material that passes itself off as reality in an informational context. Below, we break the myths apart and show how graphics and text are actually labeled today.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Where did the myth that every AI image needs a label come from?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The myth that every AI image needs a label comes from an oversimplification of Article 50 of the AI Act. The transparency obligation covers content that could mislead the viewer about whether they&#8217;re looking at real or generated material. That&#8217;s a very different thing from \u201canything AI ever touched.\u201d Technical assistance, design work and authorial work stay outside the scope of visible on-frame labeling.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Redrawing your own screenshot as a vector \u2013 by hand or with the help of an AI tool \u2013 is authorial work or technical assistance, and the AI Act doesn&#8217;t cover it. Any technical metadata (such as C2PA) gets embedded by the software, for example Adobe, not by the creator. Day to day, this \u201cno visible marker\u201d group typically includes:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>icons, logos and vector illustrations assembled in an editor,<\/li>\n\n<li>UI mockups and diagrams traced from your own screens,<\/li>\n\n<li>AI-assisted retouching, cropping and color correction,<\/li>\n\n<li>decorative graphics that don&#8217;t pass themselves off as real photography.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">What does the AI Act actually say about labeling?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The AI Act says less about labeling than the headlines suggest. To work out whether a visible on-frame marker is needed, just run through four questions in order:<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>Is the material photorealistic and does it imitate real-world reality?<\/li>\n\n<li>Does it appear in an informational context (news, journalism, an official statement)?<\/li>\n\n<li>Could a viewer mistake it for a genuine photo or recording?<\/li>\n\n<li>Only if you answer \u201cyes\u201d to all of them do you actually need a visible label.<\/li>\n<\/ol>\n\n\n\n<p class=\"wp-block-paragraph\">The classic case that does require labeling is a deepfake presented as genuine material. A generated \u201cphotograph\u201d of an event, or a fabricated statement from a politician published on a news site, needs a marker. It protects the audience from disinformation.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">How are images and text technically labeled?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Technically, images and text are labeled in two complementary ways. The first is invisible metadata embedded in the file; the second is watermarks blended into the content itself. Below, we break down both.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What is C2PA and Content Credentials?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">C2PA (Content Credentials) is a standard that attaches a cryptographically signed \u201cmanifest\u201d of the creation history to a file. It records which tool created or edited the image and whether a generative model was used, and the signature makes it possible to detect any tampering with the metadata. This layer is invisible to the eye \u2013 it lives in the file, not on the frame. In most tools (Adobe Firefly, cameras with CAI support), the software adds it automatically.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">How do SynthID watermarks and text watermarking work?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Watermarks are blended into the content itself, alongside the metadata. In images and video (for example SynthID), pixels are modified in a way that&#8217;s invisible to a human but detectable by machine. In text, watermarking works by subtly steering the model&#8217;s word choices. That leaves a statistical \u201csignature\u201d a detector can read, without changing the meaning of the sentence for the reader.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Does animating a background (water, air, a cinemagraph) require a marker?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Animating a background \u2013 rippling water, shimmering air, or a subtle cinemagraph on a banner \u2013 doesn&#8217;t need a marker when it&#8217;s used for decorative or marketing purposes. A banner with an animated background on an offer page gets by without a visible label, and a cinemagraph in a product ad carries the information, at most, in the file&#8217;s metadata. The obligation only kicks in once that kind of animation passes itself off as genuine footage in a journalistic context, for example as supposed \u201cfootage from the scene.\u201d<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Do digital actors in a film need to be labeled?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Digital actors and AI-animated characters in a film don&#8217;t need to be labeled on screen. Artistic and fictional works don&#8217;t need to be \u201cdisfigured\u201d with on-frame watermarks, because they fall under the exception for creative works.<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p class=\"wp-block-paragraph\">When content is part of a work or program that is clearly artistic, creative or fictional, an appropriate disclosure in a form that doesn&#8217;t interfere with viewing is enough \u2013 for example, in the closing credits or the film&#8217;s description.<\/p>\n<cite>Gist of Article 50(4) AI Act (paraphrased)<\/cite><\/blockquote>\n\n\n\n<h2 class=\"wp-block-heading\">When is a visible marker on an image actually needed?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">A visible marker on an image is only needed when the material passes itself off as reality in an informational context. Here&#8217;s how that plays out in typical work situations:<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table><thead><tr><th>Work scenario<\/th><th>Visible marker on the image?<\/th><th>Where to put the disclosure?<\/th><\/tr><\/thead><tbody><tr><td>A vector made from a screenshot of your own system<\/td><td>No<\/td><td>No requirement<\/td><\/tr><tr><td>Animated water or air (web banner)<\/td><td>No<\/td><td>File metadata (added by the AI tool)<\/td><\/tr><tr><td>AI-animated characters in a film or scene<\/td><td>No<\/td><td>Closing credits \/ film description<\/td><\/tr><tr><td>Photorealistic deepfake (informational)<\/td><td>Yes<\/td><td>A visible label next to the material<\/td><\/tr><\/tbody><\/table><figcaption class=\"wp-element-caption\">Rule of thumb: a visible marker belongs wherever material passes itself off as reality in an informational context.<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">So the focus falls on context, not on the tool. It isn&#8217;t the use of AI that triggers the obligation, but whether the viewer could be misled. The very same generated image can be an illustration with no requirements one time, and a deepfake that needs a label the next.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">How is the AI Act different from copyright law?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The AI Act and copyright law are two separate obligations that are easy to mix up. AI transparency answers the question of whether the viewer knows what they&#8217;re looking at. Copyright law answers the question of whether you&#8217;re unlawfully copying someone else&#8217;s interface or work. You can satisfy one obligation and still violate the other.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>A visible \u201cGenerated by AI\u201d label applies almost exclusively to photorealistic deepfakes used in an informational context. Find out when the law actually requires labeling \u2013 and how C2PA and watermarks work.<\/p>\n","protected":false},"author":1,"featured_media":905970,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_beeclear_ctt_text":"","_beeclear_ctt_class":"","_beeclear_ctt_element":"p","_beeclear_ctt_is_custom_code":false,"_ca_focus_phrase":"","footnotes":""},"categories":[449],"tags":[],"class_list":["post-1084","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-marketing"],"_links":{"self":[{"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/posts\/1084","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/comments?post=1084"}],"version-history":[{"count":0,"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/posts\/1084\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/media\/905970"}],"wp:attachment":[{"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/media?parent=1084"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/categories?post=1084"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/beeclear.pl\/en\/wp-json\/wp\/v2\/tags?post=1084"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}