Deepfakes, fake video or audio content using artificial intelligence, are increasingly present and real. They can be used for malicious purposes such as identity theft, phishing or online scams. With the development of generational AI tools like DALL-E, Midjourney or Sora from OpenAI, anyone can now create fake content in a few clicks. How to protect yourself from it? Here are 7 tips for spotting deepfakes.
Look at the light and shadows in the image
The first reflex is to have itcarefully examine lights and shadows on a suspicious photo or video. With deepfakes, the main subject is often sharp and realistic, while the background may lack detail or look artificial.
So compare the light on the face with the ambient light. Are there inconsistencies? A shadow that doesn’t match the position of the sun? This can be a sign of manipulation.
Move into face and mouth
Another tip: move in on the the person’s face, especially around the mouth and eyes. Make sure the skin tone is fair between the face and the rest of the body. See if the facial expressions are clear or on the other hand vaguely and badly cut out.
For video, note if the lip movements are perfectly synchronized with the noise. You can also focus on the teeth: are they sharp or dull? according to Nortonalgorithms are still struggling to realistically generate teeth.
Examine the full context
Take a step back and ask yourself about the plausibility of the content in terms of the context. Does the situation seem credible and plausible to you? As the site advises Poynterbe wary if a public figure’s behavior seems inconsistent or complex compared to the norm.
“If the Pope really did wear a luxury jacket, wouldn’t other pictures have been published by legitimate sources?” »showing theAssociated Press. So check if there is any additional shots coming from official media.
Use AI search tools
To identify deepfakes, you can also rely on… artificial intelligence! Open AI for example, it’s just a special tool to create images with FROM E 3.
For his part, Microsoft has developed Video checkerable toanalyze a video or photo and identify the probability that it is fake. The chip maker Intel also offers its detector FakeCatcher.
Be careful with the tools
Be careful, however, not to blindly trust these detection technologies.
“It would be dangerous to imagine them to be perfect machines that can be completely trusted”giving a warning Henry Ajderexpert in generative AI.
It is necessary to keep them in mind Bordersespecially since AI is advancing at a rapid pace.
“There is no guarantee that this advice will be valid in a year”he emphasizes it.
Finding will become more difficult, even to a trained eye.
Check the source and the facts
More than ever, it is essential information cross-check, especially when content seems questionable to you. Look where it came from, go back to the sources of origin if possible. look for fact check articles on the subject.
Talk to the reference media please bring up this viral image or video as well. If they don’t talk about it even if it’s all the rage, it’s probably fake. Research before you share!
Train your critical thinking against deepfakes
With rapid advances in AI, even search engines are struggling to keep up. To protect ourselves, we still have the best weapon surveillance and our a critical mind.
“We must not rely entirely on technology and think that it will think for us”wants Henry Ajder.
Let’s stay alert for inconsistencies, check the facts, and consult several reliable sources. This is the key to do not fall into the trap of the deep.
My name is Ethan, I am 30 years old, and I am the founder of this magazine. After studying journalism, I quickly decided to go solo to be able to write the way I wanted and talk about my true interests. I aim to create an information center where French and international news are treated seriously and with insight. My aim is to enlighten the daily lives of my readers and promote a deep understanding of current affairs.
