Brazilian AI Detects Deepfakes: A Cat-and-Mouse Game with Technology

Brazilian Researchers Develop AI to Combat Deepfakes Brazil is at the forefront of the fight against deepfakes, thanks to innovative AI technology developed at the University of Campinas (Unicamp). This cutting-edge system analyzes videos in stages, focusing initially on facial recognition to detect inconsistencies indicative of AI manipulation. The technology is already being utilized by the Public Ministry in São Paulo and Bahia to investigate cases involving potentially fabricated video evidence. "Today, if you look at a video or image generated by artificial intelligence, most likely you, as a human, will not identify that it is false. You really need another artificial intelligence to identify," explains Anderson Rocha, a professor at Unicamp's Institute of Computing. Gabriel Bertocco, another researcher involved in the project, details the AI's multi-stage process. "Once the face is detected, it performs the analysis. The other frames are processes of analysis of the model." However, the researchers acknowledge the ongoing challenge of keeping the detection technology updated to counter the rapid advancements in deepfake generation techniques. They describe the situation as a continuous "cat and mouse game." This development underscores the critical role of AI in combating the spread of misinformation and ensuring the integrity of digital information. The ongoing refinement of this technology offers hope in the fight against deepfakes and the protection of public trust in online content.