Astronomers’ Galaxy-Studying Techniques Are Helping To Identify AI-Generated Deepfakes

The discovery that AI-generated deepfakes can be identified by analyzing the reflections in people’s eyes, similar to how astronomers study galaxies, has significant implications for combating the spread of misinformation.

Key findings: Researchers at the University of Hull have developed a method to detect AI-generated deepfakes by examining the consistency of light reflections in a person’s eyeballs:

• In real images, the reflections in both eyeballs are generally consistent, while in deepfakes, the reflections often lack consistency between the left and right eyes.
• By employing techniques used in astronomy to quantify the reflections and check for consistency, the team found that this method can effectively distinguish between real images and AI-generated fakes.

Adapting astronomical techniques: The researchers applied methods typically used by astronomers to analyze the light distribution in galaxies to the problem of identifying deepfakes:

• The Gini coefficient, normally used to measure how light is distributed among pixels in a galaxy image, was employed to compare the similarity of reflections between left and right eyeballs.
• While the CAS (concentration, asymmetry, smoothness) parameters, another tool used by astronomers to determine galaxy morphology, were tested, they were found to be less effective in detecting fake eyes.

Limitations and future implications: While this method provides a promising approach to identifying deepfakes, it is not a perfect solution and may produce false positives and negatives:

• Professor Kevin Pimbblet, director of the Centre of Excellence for Data Science, Artificial Intelligence and Modelling at the University of Hull, emphasized that this technique is not a “silver bullet” but rather a basis for attacking the problem of deepfakes.
• As AI-generated images become increasingly accessible and sophisticated, the ability to detect fakes, particularly of people, is becoming more critical in the fight against misinformation.

Broader context: The research, conducted by MSc student Adejumoke Owolabi, highlights the innovative application of techniques from one scientific field to another, demonstrating the potential for cross-disciplinary collaboration in tackling emerging challenges posed by artificial intelligence. As deepfakes continue to evolve and become more difficult to identify, this study underscores the need for ongoing research and development of detection methods to stay ahead in the “arms race” against misinformation.