It may be getting harder to pick a fake, but there are tools to help

There was a time when you could look at a photo and be fairly sure it showed something that really happened. But with the rise of artificial intelligence (AI), that is no longer always the case. Some images, videos and sounds are created or changed by AI to look real when they are not.

These fake but realistic creations are called deepfakes. According to the eSafety Commissioner – a government organisation that helps keep people safe online – deepfakes are a type of synthetic* media, where AI creates or changes videos, audio or images to make it appear that a real person is doing or saying something they never actually did.

The word “deepfake” comes from combining the words “deep learning” and “fake”. Deep learning is a type of AI that helps computers learn from large amounts of information.

Curtin University Professor of Internet Studies Tama Leaver said technology was changing very quickly. He said it was important to understand how widespread the problem was and to think carefully before believing everything you see or read online.

“‘Deepfakes’ is quite a big term now,” Prof Leaver said. “Initially it was the relatively rare occurrence of an image being manipulated* in a way that simply told a different story. Early deepfakes were faces being swapped out, so you’d have someone doing something with a different face over it. Now AI apps are able to generate any imaginable scenario and put in real images of people just by seeing your photo.

“The real challenge is a photo doesn’t purport* to tell the truth anymore. It makes it very difficult to tell when all sorts of things are and aren’t true.”

Some AI tools, like Google Gemini, add a sparkling star logo to the corner of AI-made pictures, so people know the images aren’t real photos. But many other AI tools don’t add any clear sign, sometimes called a watermark, which can make it harder to tell whether an image was created by AI or not.

In the past, it was often easy to spot a deepfake, because the technology wasn’t very good. Pictures might have had strange backgrounds, people with the wrong number of fingers, or other funny-looking mistakes. But the technology has improved so much that many deepfake images now look just like real photos.

And they are likely to get even better as time goes on.

Prof Leaver said when it comes to identifying deepfakes, it was important to ask yourself three questions: Who is sharing the image? Why was it made? Do you have reason to trust the person or institution sharing it?

Government-funded news organisations, such as the ABC and SBS, are often seen as trustworthy because they must carefully check that information is accurate* and true before they share it with the public.

Many other trusted news organisations (including News Corp Australia, publisher of Kids News, and international news agencies like Reuters) also carefully check their facts before publishing stories. However, more and more people now get their news from social media platforms such as TikTok and Instagram, where posts are not always checked in the same way by news organisations or professional editors.

Dr Maša Popovac from Australian charity the Alannah and Madeline Foundation, said if something makes you feel surprised, excited or angry, it’s usually best to pause before sharing.

“AI content is often created to get a reaction, so check who posted it and think about why they may have done so,” she said. “You can also look for other evidence. If something really happened, then other sources are likely to be talking about it too.”

There are also reverse image search tools, such as Google Lens or Bing Visual Search, that allow users to see where a picture first appeared online. Finding out the origin could help you discover if an image was changed or shared with a false story.

Another smart step is to check trusted fact-checking websites, including Snopes, AFP Fact Check and Reuters Fact Check. These organisations investigate claims, photos and videos to help people tell the difference between what’s real and what’s fake.

The best way to stay one step ahead of deepfakes is to stay curious, ask questions and think before you trust or share what you see online.

POLL

GLOSSARY

synthetic: produced artificially through chemical processes rather than occurring naturally

manipulated: to control or influence something or someone

purport: to claim something is true

accurate: free from error or correct in all details

EXTRA READING

Part 1: Your digital wellbeing

Part 2: Your news filter

Part 4: Clickbait and engagement traps

QUICK QUIZ

What three things can AI change to create a deepfake?

How did they come up with the word ‘deepfake’?

What three questions should you ask yourself when trying to identify a deepfake?

Name two reasons why it’s getting harder to identify a deepfake

What tools can you use to help identify a deepfake?

LISTEN TO THIS STORY

CLASSROOM ACTIVITIES

Please refer to the accompanying Digital Literacy classroom workbook with 24 teacher-produced, curriculum aligned activities. It’s FREE for teachers who have signed up to receive the weekly Kids News edm.