More and more deepfakes are circulating on the Internet. They pose a danger because they appear deceptively real. A new study now provides a method to distinguish deepfakes from real images.
Deepfakes are deceptively real-looking but artificially created or altered photos, video or voice recordings. That’s the German definition Federal Government.
The problem: Cybercriminals use deepfakes for phishing, disinformation and manipulating public opinion. Because of advanced artificial intelligence, it is now very difficult to expose deepfakes.
Study explains how deepfakes differ from real recordings
A study by the Royal Astronomical Society shows that there is one detail in many deepfakes that distinguishes them from real photos: the reflection of light in the eye. AI-generated counterfeits can be detected by analyzing the human eye.
In one official press release The astronomers show an image as an example: On the left you can see a real image of the actress Scarlett Johansson. On the right is a photo of a person created by an artificial intelligence.
Below the image are close-ups of both people’s eyeballs. There you can see that the reflections are consistent for the real person and physically wrong for the artificial person.
This means that if the reflections in the eyeballs match, the image is probably that of a real person. If they are unequal, it is probably a deepfake.
Why are astronomers studying artificial intelligence?
The study is the result of research by Adejumoke Owolabi, a master’s student at the University of Hull in Yorkshire, England. She analyzed AI-generated fakes in the same way astronomers examine images of galaxies.
“To measure the shape of galaxies, we analyze whether they are centrally compact, whether they are symmetrical and how smooth they are. We analyze the distribution of light,” explains Kevin Pimbblet, professor of astrophysics and director of the Center of Excellence for Data Science, Artificial Intelligence and Modeling. “We detect the reflections in an automatic way and run their morphological features through the CAS and Gini indices.”
Researchers typically use the Gini coefficient to measure how light is distributed across its pixels in an image of a galaxy. To do this, they are arranged in ascending order according to their luminous flux and then compared with the result that would be expected with a completely uniform luminous flux distribution. People’s left and right eyeballs can be compared in the same way.
The scientific method used by the University of Hull team is not a panacea for detecting fake images, says Kevin Pimbblet. There would also be false positives and false negatives. However, the approach provides “a plan of attack for the arms race in detecting counterfeits.”
Also interesting:
- Deepfakes: Bavaria demands a penalty of up to 5 years in prison
- Threatening EU penalties: Tech companies must take action against deepfakes
- CT scans: Artificial intelligence could save thousands of lives
- Zombie Brands: AI Websites Steal Authors’ Identity
The article Study reveals: This is how you can distinguish deepfakes from real recordings by Beatrice Bode appeared first on BASIC thinking. Follow us too Facebook, Twitter and Instagram.
As a Tech Industry expert, I find the findings of this study to be incredibly important in the ongoing battle against deepfakes. Being able to distinguish between deepfake and real recordings is crucial in order to prevent misinformation and manipulation in various fields such as politics, journalism, and entertainment.
The study’s identification of key visual and audio cues that can help detect deepfakes is a significant step forward in improving our ability to spot these deceptive videos. By understanding the telltale signs of manipulation, we can develop more effective tools and strategies for detecting and combating deepfakes.
It is essential for tech companies, researchers, and policymakers to continue investing in research and development efforts to stay ahead of the rapidly evolving technology used to create deepfakes. By working together and sharing knowledge, we can better protect the integrity of information and ensure that trust and authenticity are maintained in our increasingly digital world.
Credits