With algorithms changing constantly and new rules popping up left, right and centre, it can be an actual nightmare trying to figure out which ones are actually worth your attention. As a marketing agency that prides ourselves on utilising the latest AI functionalities to streamline and enhance our workflows, from assisting with competitor analysis to analysing large data sets of results and even using the handy Canva tools like background remover occasionally, we thought it was only right that we introduce you to this law.
The EU’s new AI transparency law can seem complicated at first, but once broken down, it’s actually much simpler than it sounds. In this guide, we’ll dive into what the new act is, what has changed, who it affects, and a few tips to help your business stay compliant through this change.
These new rules are part of the EU AI Act. You might have seen it be called the “AI Transparency Rule” or “AI Visibility Rules” online, but it’s all the same thing, so don’t let the different names throw you off. The main purpose of this new act is to ensure that people now have the right to know when they’re talking to AI, or when what they’re looking at was made or altered by AI.
And that is pretty much it, no need for complicated jargon. This new act isn’t a licensing scheme; it doesn’t ban the use of AI tools, and it also isn’t aimed at the technology itself. It’s also not to be confused with the AI Act’s “high-risk AI” rules, which target things like AI used in hiring decisions or medical devices.
The EU AI Visibility Act (Article 50) has been in force since 2 August 2026, and it covers businesses using generative AI to create content. It can also apply to businesses outside of the EU, meaning that if your AI-generated content reaches people within the EU, these rules may apply to you too. This isn’t just an EU-based business issue, so it’s worth understanding what the rules mean for your business. The good news is there are some clear guidelines around when disclosure is needed, so you don’t need to panic about every piece of content you create with assistance from AI.
If you’re running an AI chatbot on your website, using AI to write social captions/ blog posts, or generating product images or videos with tools like Sora, this law could affect you, purely because some AI content now needs labelling as AI.
On top of this, deepfake-style content is getting extra attention – think AI voice, AI video, or realistic AI images of real-looking people. That’s treated more seriously than everyday AI-assisted text or graphics, because the chance for confusion or manipulation is much higher. So it’s worth knowing which section your content falls under: putting your product into AI and generating images of it in different settings is a very different situation from an AI-generated video of someone who looks like a real person talking about their experience using your product, which in reality is a cleverly crafted script designed to deceive and mislead consumers.
Article 50 comes down to four specific scenarios. That might sound like a lot, but most everyday marketing teams will only need to focus on a couple. Here’s what they mean in practice:
- People interacting with an AI system: If someone chats with a bot on your site, they need to know it’s a bot, not a real person. This could be as simple as making it clear at the start of the chat that the conversation is with an AI assistant. You should also consider that many consumers are anti-AI, so providing an easy way for people to bypass the chatbot can ensure better customer service and less frustration.
- AI-generated audio or video that resembles real people or events: This is where deepfakes come in. If AI is used to create or manipulate audio, video, or images that could make people believe they’re seeing or hearing something real, it must be clearly disclosed. You can’t use AI to make a video of a celebrity endorsing your product, or claim you put solar panels on Everest – it sounds obvious, and we all know it’s not acceptable, but unfortunately it has become an issue, which this new legislation will hopefully prevent.
- AI-generated text on matters of public interest, published without human editorial review: If AI-generated text covers something that could significantly impact the public, such as politics, health, or current affairs, and no human has reviewed it, it needs to be disclosed as AI-generated. This is one area where human review becomes particularly important. AI can and often does tell lies; it can make things up, use incorrect figures and data, and unless you have a human editing and auditing, you could be accidentally miseducating your community.
- Emotional recognition or biometric categorisation systems: This is a much narrower part of the rules and probably isn’t something most marketing teams will come across in their everyday work. It covers AI systems that analyse things like someone’s emotions and use biometric information to categorise people. If you’re simply using AI to write a social post caption, then this isn’t the part you need to worry about.
There are a few exemptions in the rules that make things seem much less scary than they might first sound:
A human has checked it: If a real person has reviewed the content, made changes where needed, and had the final say on what gets published, it can often fall outside the strict disclosure rules.
You’re using AI to tidy things up: Using AI for things like SPaG or small edits doesn’t immediately mean it needs to be labelled as AI, just as long as it hasn’t fully changed your original words.
It was published on 2 August 2026: Please don’t start panic-digging through old, published blogs and social posts labelling everything as AI! Anything published before this date doesn’t need to be labelled retrospectively.
Overall, if your usual process is to use AI as a little helping hand and then have a human give everything a complete look-through before it goes live, you’re already complying with these rules.
No need to change everything overnight. Start by taking a proper look at where AI actually affects your content pipeline, whether that’s copy, images, video, or a chatbot on your site. Once you know where AI is being used, take a look at anything that’s gone out with little to no human review. That’s usually the first thing to sort.
From there, build a few small habits into how you already work. Add clear, simple disclosure where needed; something straightforward like “created with AI assistance” works fine. It’s also worth noting that many social media channels, like Instagram, for example, will already label AI content on their platform, so you don’t need to fret about adding notices to every post. Review your contracts with any AI tool vendors too; some providers already handle parts of the compliance side for you. And before anything AI-assisted goes live, add a quick human sign-off step. We also suggest keeping a simple record of what’s been checked, so you have something to point to if anyone ever asks.
Now the part that nobody likes talking about: what happens if you just don’t bother. Fines can reach a whopping €15 million or 3% of global turnover. That’s a serious number, and it’s meant to be, but this law isn’t designed to catch people out for an honest mistake; it targets businesses that ignore the rules altogether, not those making a genuine effort.
Now this is the real question. AI makes it very easy to create marketing content – a few minutes of entering well-written prompts could produce enough social captions for a month. But should you use it that way? Well, we are big advocates for AI, but we also know that it is a fantastic tool and not a complete marketing solution. AI can only use what is already out there – it digs through the internet and can compile plenty of data, but it cannot be creative. It cannot understand human emotions, thought processes, or feelings, and as a result, the content it creates lacks the magic touch – humanity.
Unfortunately, there are a lot of snake oil sellers out there that push AI as a simple solution to marketing, but it isn’t. It creates content that is literally designed to look exactly like your competitors’. It’s like using a graphic designer who only knows one style – and all your competitors are using them too! The content it creates rarely stands out as it is everywhere, and it can even leave a sour taste in your audience’s mouth especially as consumers are increasingly anti-AI. Awareness around the AI race, the environmental impact and the constant onslaught of AI slop has meant that many consumers are even leaving brands that utilise AI too heavily in their creatives or choosing not to engage with them at all.
We believe AI is a great way to enhance your marketing, but using it as a standalone solution isn’t the right way to harness the technology. You wouldn’t ask ChatGPT how to rewire your home and then go straight to B&Q to do it yourself; you would understand that the experience and knowledge of a trusted tradesperson far outweigh the opinion of AI – marketing should be viewed the same.
At Cheshire Cat Marketing, we love helping businesses adapt to the changing AI landscape. From ensuring your business complies with new laws and rules to helping you find the best ways to use AI to optimise your business fully, we truly believe AI can be a powerful tool when used correctly.
It doesn’t have to be complicated or scary, and with the proper approach, AI can save you time and boost your marketing for the better, and we’d love to help. Visit our website, book a discovery call, or reach out to our team to learn how we can help your business grow – without losing that important human touch.
