For many years, one of the simplest ways to check information was to look for proof. A photo was proof. A video was even better. If we could see a person saying something with our own eyes, there was usually little reason to doubt it.
Today, this is changing. Artificial intelligence can create images of people who do not exist, reproduce a person’s voice and generate videos showing events that never happened. These technologies are also becoming much easier to use. You no longer need to be a professional designer or video editor. In many cases, you just need an AI tool and a few minutes.
So what happens when seeing is no longer believing?
A fake image does not need to be perfect
When AI-generated images first became popular, people often looked for strange details: six fingers, unnatural eyes, incorrect shadows or text that made no sense. These signs were useful, but technology is improving very quickly. AI-generated images and videos are becoming much more realistic and many of the old signs are becoming less reliable.
There is also another problem. A fake does not always need to convince everyone. Imagine that you are scrolling through social media and see a shocking video. You watch it for five seconds, read the caption and move on. Maybe you do not stop to look at the hands, the light or the background. Maybe you just remember what you saw. Sometimes that is enough.
Deepfakes are not only about video
The word “deepfake” is often connected with videos where one person’s face is replaced with another. But the problem is now much wider. AI can generate photos, imitate voices and create realistic video. It can also produce large amounts of text very quickly.
Imagine receiving a voice message that sounds like someone you know or seeing a video of a public person apparently making a statement that was never made. Would you recognize that it was generated? Sometimes you might. Sometimes you might not. This is why media literacy cannot depend only on our ability to “spot the fake”.
But AI is not the only problem
There is an important point which is easy to forget: not every misleading image is created by artificial intelligence. Sometimes the photo or video is completely real.
A photograph from one protest can be presented as another protest. A video recorded several years ago can suddenly appear online with the words “This is happening now”. A short part of a longer video can remove important context and completely change how we understand what happened. In these cases, looking for strange fingers or visual mistakes will not help. The image is real, but the story around it is not.
This is why the first question should not always be “Is this image AI-generated?” A better question can be: “Where did this image come from?”
Go back to the source
When you see a surprising photo or video, try to find its origin. Who posted it first? When was it published? Where was it recorded? Is there a longer version? Are reliable news organizations or official sources reporting the same event?
This approach can be more useful than trying to become a deepfake detective. A screenshot can hide where information came from. A short video can remove what happened before and after. A repost can separate an image from its original date and location. Going back to the original source helps us rebuild this missing context.
Be careful with your emotions
There is another reason fake and misleading content can work so well: we often see it when we are moving fast. Social media encourages immediate reactions. We like, comment, share or send something to another person in a few seconds.
Content that makes us angry, frightened or surprised can make this reaction even faster. This creates a simple problem: the moment when we most want to share something may also be the moment when we most need to check it. A short pause can make a difference. Where did this come from? Who published it? Can I find the same information somewhere else?
These are simple questions, but they become more important when technology makes false content more convincing.
What does this mean for education?
Young people are growing up in an environment where creating and changing digital content is becoming normal. Telling students simply “don’t believe everything you see online” is no longer enough.
They need opportunities to work with information, compare sources, question images and discuss why something looks convincing. They also need to understand how a real image can be used in a false context. Teachers do not need to become AI experts to do this. A photo, a social media post or a short video can already become a classroom activity.
Students can try to answer simple questions. Where did the content come from? What information is missing? Can another source confirm it? What would we need to know before sharing it? These questions develop a skill which remains useful even when technology changes: critical thinking.
Seeing is only the beginning
Artificial intelligence will continue to improve. The next fake image will probably be better than the previous one. New tools will appear and some of today’s methods for recognizing generated content will become less useful. Trying to memorize every sign of an AI-generated image is therefore not enough.
We need a different habit. When something looks real, do not immediately ask yourself whether your eyes can detect the fake. Ask where it came from, who created it and what happened before and after. Look for another source before deciding what to believe.
JUMP is working on these skills through Fake News Busters, an Erasmus+ project dedicated to media and digital literacy, critical thinking and helping students and teachers better recognize misinformation.
In a digital world where images can be real, manipulated or completely generated, perhaps the most important lesson is very simple: seeing something is no longer the end of the verification. It is the beginning.
Author: Yefimnko Yevhenii (JUMP staff)