Background information
AI Act: how the EU wants to put artificial intelligence in its place
by Florian Bodoky
Misleading AI videos and what to do about them
11.2.2026
Translation: Katherine Martin
When the first wave of AI-generated content hit, the videos were easy to recognise as fakes. This is no longer the case – and we’ve got to figure out how best to deal with it. While I don’t have the solution, I do have some food for thought.
YouTube’s suddenly full of videos of old people telling their life stories. Sometimes, they’ll be talking about the dietary choices that helped them reach the age of 100. Other times, they’ll be describing the mindset their 90-year-old selves wish they’d had at 20. Personally, I find it interesting when people look back on their wealth of life experience and draw important insights from it.
So, where do all these remarkably wise elderly folks come from? Well, out of thin air. They’re not real people – they’re AI-generated. And the words they say are either written by AI or by a relatively young person with all their marbles left. Even if the «inspiration» behind the script really did come from elderly people.
Fiction versus fakery
This annoys me. Why? Well, we had made-up biographies in the past too, for example in novels. But there’s an important difference there. If a story’s labelled as a novel, you know from the outset that the story’s fictional. You’re not led to believe that it’s a real story about a real person.
In these videos, that’s not the case. Hidden somewhere in the lengthy, pop-out description, there’s a note that the person’s AI-generated. The words at the top of the page, however, suggest the opposite. For instance: «We are looking for real stories from real people to feature on this channel. If you have a life lesson, a deep regret, or a near-death experience that changed everything, we want to hear it.»
Reading the comments under the videos, it’s evident the viewers haven’t noticed this well-hidden note. They address the non-existent person directly, thank them and wish them all the best.
It’s in the creators’ interest that things stay that way. After all, the entire credibility of these videos is based on the people depicted in them being real. The main merit of a particular diet is that somebody has lived over 100 years by following it. The convincing, poignant thing about these life stories is that, when faced with their mortality, the person clearly sees what life is – or would’ve been – all about. If I then find out that this person never lived, I feel cheated. Even if the advice itself is good or expressed similarly by real people, the videos are still taking viewers for a ride.
The irony shoots up to sky-high levels when an AI-generated person tells me I’m deceiving myself about something. There’s only one person trying to deceive me, and that’s the creator of the video!
We have a fundamental problem on our hands
However, these made-up seniors are a fairly harmless example of AI. The substance of what they’re saying isn’t bad. In fact, it might actually help some people. The thing is, AI videos that look real can also be used for malicious purposes, including spreading fake news and targeted disinformation. I don’t think I need to elaborate on why the world’s already in dangerous territory when it comes to these issues. With AI videos, disinformation trolls have another very powerful tool at their disposal.
Now, you might be thinking that you wouldn’t fall for something like that. I think you’re wrong. Sure, if you know a video isn’t genuine before watching it, you’ll see signs of fakery all the way through it. For instance, that a woman over 90 can speak freely and uninterrupted for so long, without showing any signs of tiredness. That her remarks sound so polished. Even the strange sense of consistency running through everything she says.
But there are always moments when our inner sceptic fails to wake up. It’s like with phishing e-mails. You can spot them 100 times, but eventually you’ll still fall for one. If your suspicions aren’t aroused when watching a video because there’s seemingly no obvious reason for it, plenty of AI videos are already good enough to fool us all.
In future, these videos will be even more realistic. Just think about how much they’ve evolved over the last three years. We’ll soon have AI videos that the average viewer won’t be able to distinguish from genuine videos. And rather than just a handful of people, it’ll be millions.
What do we do now?
The question is: how are we going to deal with this? A response is required at the political, legal and economic levels. As individuals, we also need to think about how we’ll deal with the problem ourselves.
Politicians need to regulate the use of AI videos. The EU has taken an important step on this by introducing an AI Regulation, which divides AI applications into different risk categories. Deepfakes aren’t classified as high-risk; regulation’s limited to mandatory content labelling.
On Tiktok, Instagram or YouTube Shorts, AI disclaimers buried deep in a description don’t do any good. Nobody sees them. The rules governing what AI labelling should look like still need to be defined more precisely.
What can you do as a small, powerless user in the here and now? Scepticism’s good, but it definitely shouldn’t go so far that you stop believing absolutely everything. Rather than solving the problem, that makes everything worse. Not only are people who doubt everything as a matter of principle seriously tedious, but they also destroy meaningful discussion. No longer convinced by anything, they cast doubt on the foundational elements of our civilisation. Although quite a few people like that already exist, I don’t want to turn into one of them.
Here’s what I’m planning to do: no more mindlessly scrolling through videos or clicking on random stuff on YouTube. In future, I’ll be selective about the videos I consume and, more importantly, their creators. I’ll also give them my full attention as I’m watching (and keep my brain switched on). In this day and age, being able to confidently decide who or what to pay attention to seems to be one of the most important skills you can have.
Since social media tends to just throw stuff together, I’m planning to go back to consuming more traditional media again. That way, I’ll have a better idea of where something’s come from and what the context is. It’s also good to let people know when they’ve been hoodwinked by AI. But only if you’re absolutely sure. As I said, being suspicious of everything and everyone isn’t helpful.
Header image: all the old people in the world who have ever been filmed
David Lee
Senior Editor
David.Lee@digitecgalaxus.chMy interest in IT and writing landed me in tech journalism early on (2000). I want to know how we can use technology without being used. Outside of the office, I’m a keen musician who makes up for lacking talent with excessive enthusiasm.
Photo and video
Follow topics and stay updated on your areas of interest
Computing
Follow topics and stay updated on your areas of interest
Opinion
This is a subjective opinion of the editorial team. It doesn't necessarily reflect the position of the company.
Show allThese articles might also interest you
98 comments
later