In recent months, AI-generated photos have sparked a wave of misinformation online in areas ranging from fake advertising to art theft.
New York Times reported that the image recognition software created with artificial intelligence is one of the solutions that can be used to prevent the spread of false information produced with artificial intelligence; Now this publication has announced that it is possible to deceive artificial intelligence image recognition systems with a very simple method.
A New York Times analysis shows that when you add texture to an AI-generated photo, the chance that software will recognize it as an AI-generated photo drops from 99 percent to 3.3 percent. This method even works on Hive, one of the best image recognition software made with artificial intelligence, and you can trick this program by applying textures to photos.
Cynthia Rudina professor of computer science and engineering at Duke University, said in an interview with the New York Times: “As the use of artificial intelligence-based image generation tools expands, better software will be developed to recognize these types of images.”
The New York Times analysis comes as users increasingly abuse artificial intelligence-generated misinformation online.