The EU’s new AI Act explained

What this article isn’t about

Whenever AI is mentioned, somebody will inevitably raise concerns about data centres, electricity consumption, water usage, copyright, taxation, employment or whether artificial intelligence will eventually become self-aware and shoot a puppy. These are all perfectly valid discussions, nobody here will dispute that. However, they are also completely different discussions.

This article is about one specific part of the AI Act: the transparency rules surrounding AI-generated and AI-manipulated content. The wider debates deserve proper consideration in their own right, rather than being squeezed into an article that’s trying to answer a different question. That’s one for another day.

What is the AI Act?

If you would like to read it for yourself, you can click HERE.

The AI Act is the European Union’s first comprehensive law governing artificial intelligence. Contrary to what some corners of Facebook would have you believe, Brussels hasn’t banned ChatGPT, declared war on Canva or dispatched inspectors to count how many toes your AI-generated Labrador has.

The aim is actually rather sensible. As AI becomes more capable of creating realistic text, images, video and audio, the EU wants to make sure it’s used responsibly and that people aren’t easily deceived by content that looks genuine but isn’t.

Unfortunately, somewhere between the legislation being published and social media getting hold of it, the message became:

“Everything created with AI must now have a disclaimer.”

It’s a cracking headline but it’s simply not an accurate one.

So do I have to disclose AI-generated content according to the EU’s new AI Act?

For the vast majority of people, no. If you use AI to help write an article, create a Social Media graphic or tidy up a paragraph that stubbornly refuses to make sense, you haven’t suddenly wandered into a legal minefield. You can safely ignore the trolls who tell you different (ironically often using AI themselves because they’re incapable of typing words with multiple syllables… or are even a bot)

The EU’s new AI Act isn’t sitting there waiting to pounce because your local café asked ChatGPT to suggest a caption for tomorrow’s cake photo. It’s far more interested in situations where AI is used to convince people that something fake is actually real. That’s a vital distinction.

AI is simply another tool. Before AI came along we had Photoshop. Before Photoshop we had darkrooms. Before that people airbrushed things by hand. Humans have been finding inventive ways to improve pictures for centuries. AI just happens to be the latest shiny new toy.

What about social media?

This is probably where the greatest amount of unnecessary panic has appeared. 

• Imagine you own a bar in Spain and you’ve created a smart-looking graphic announcing your new tapas menu. Perhaps you’ve used AI to produce the background, choose the colours or make everything look a bit more polished than your own artistic abilities would otherwise allow.
• Or maybe your local hairdresser wants to announce new opening hours.
• The florist has a Valentine’s promotion.
• The café has started serving breakfasts.
• The village shop is staying open on Sundays.
• The Ayuntamiento has a slip ‘n’ slide this weekend

Do those businesses suddenly need to plaster *”Made with AI”* across every post? In most cases, no.

They’re advertising genuine products, genuine services and genuine opening hours. The graphic itself isn’t pretending to be documentary evidence of anything. It’s simply doing what advertising has always done: grabbing your attention long enough to tell you something useful.

Of course, someone will almost certainly appear in the comments declaring:

“That’s AI!!!”

Congratulations to them for solving the mystery. We’d never have guessed that the cartoon tomato with a beaming smile and a hat on wasn’t photographed in real life.

Using AI doesn’t automatically make a business dishonest. It simply means they’ve used another tool to help produce their marketing.

What about holiday rental listings?

Now we’re entering much more interesting territory. Suppose a holiday rental advert shows a spectacular sea view that doesn’t actually exist. The swimming pool has mysteriously doubled in size. The garden now resembles something from a high-end villa in Costa del Luxury rather than a two-bedroom apartment outside Torre los Doofus. The sea view has morphed from a “stand on a chair, relying on a bungee strap to stop you falling and squint hard” into a “180 view complete with a spectacular setting sun, even though it’s facing north-east”. 

Whether those images were generated entirely by AI or simply edited until reality packed its bags and left almost doesn’t matter.

The issue isn’t the software or tools used. The issue is whether somebody is being encouraged to book a property based on an impression that simply isn’t true. The same applies to hotels, estate agents, restaurants and virtually every other business advertising products or services. If your restaurant is advertising thrice-cooked chunky chips, you shouldn’t be serving up fries. 

AI isn’t a magical exemption from honest advertising.

What about political images and videos?

This is where the legislation becomes far more understandable.

AI is now capable of producing remarkably convincing images showing protests, disasters, crime scenes or public figures doing things that never actually happened. Shared often enough, those images can influence public opinion before anyone has had time to ask awkward questions like, “Is this actually real?”

Strangely enough, people deliberately creating misleading political propaganda rarely stop to add a helpful caption saying, “Please note that today’s fake outrage was generated entirely by artificial intelligence.” That’s rather why these transparency rules exist.

Instead, much of the responsibility falls on social media platforms to identify manipulated content, apply appropriate warnings where possible and reduce the spread of material that’s designed to deceive.

Will every fake image disappear overnight? No. But making it more difficult for fabricated content to masquerade as genuine is a fairly reasonable place to start.

What about bloggers and publishers?

This is where things become much more relevant to websites like Spain Buddy. Many publishers now use AI somewhere within their editorial process. It might help organise research, suggest a better headline, improve grammar, summarise background information or produce a rough first draft that bears only a passing resemblance to the finished article (like this very article).

That doesn’t automatically mean every article now requires an AI disclaimer. The important point is editorial responsibility. If a human reviews the content, checks the facts, edits the article, rewrites where appropriate and ultimately decides whether it should be published, responsibility still rests with the publisher. That’s very different from pressing a button and publishing whatever the machine happens to produce.

How Spain Buddy uses AI

Since this article is all about transparency, it would be rather hypocritical not to explain our own approach.

Yes, Spain Buddy uses AI. It helps us research topics, organise information, improve structure, proofread articles and occasionally create original illustrations. Sometimes it saves us hours. Sometimes it produces absolute drivel that disappears into the digital equivalent of the recycling bin. What it never does is replace editorial judgement.

Every single article published on Spain Buddy is reviewed, edited, fact-checked and approved by a human editor before it appears on the website. Editorial responsibility remains entirely ours.

You also won’t find a little AI disclaimer stuck beneath every article. That isn’t because we’re trying to hide anything. Quite the opposite. Instead, we’ve chosen to include an AI section in our Code of Ethics explaining exactly how AI fits into our editorial process. Personally, we think that’s considerably more useful than adding the same one-line disclaimer to hundreds of articles until everyone automatically scrolls past it without reading. Christ – we’ve already got enough going on underneath our articles without it turning into an essay.

Are you still with us?

If you’ve reached this point hoping to discover whether you now need to add an AI disclaimer beneath every Facebook post, every blog article or every promotional graphic, the answer is often no. The AI Act isn’t really about whether AI was involved. It’s about whether people are being deceived.

For the vast majority of cafés, bars, restaurants, shops, bloggers and small businesses, AI is simply another tool. Used responsibly, it can save time, reduce costs and produce better content.

Used irresponsibly, it can create fake evidence, spread misinformation and persuade people to believe things that never happened. That’s what these new rules are trying to tackle. Not your Wednesday lunchtime special.