Ant Group’s Tech Brand Launched “ZOLOZ Deeper” to Combat Deepfake Threats

On April 14th, Ant Group’s tech subsidiary, ZOLOZ, launched a comprehensive solution to combat Deepfake threats, named Deeper. This product is designed to safeguard users from the risk of AI face swapping during facial recognition processes. ZOLOZ, known for its expertise in biometric technology, has proven its capabilities in large-scale applications such as the Alipay App.

ZOLOZ Deeper employs an integrated technology system to handle Deepfake security issues across systems, servers, and applications. At the user interface level, it utilizes precise sensor checks and dynamic risk control based on multi-dimensional features. This allows for strict scrutiny of factors like camera status, device and network conditions, and input images, ensuring a secure software and hardware environment and the integrity of collected biometric data. The product also uses deep learning models to analyze user facial movements meticulously, capturing subtle details like micro-expressions, muscle textures, and eye movements. This helps accurately determine whether the user is a real person interacting in real-time.

The Tianji Lab of Ant Group supports ZOLOZ Deeper’s development by generating over 300,000 test samples using the GAN model. These samples are used for discriminative training. The product also undergoes more than 20,000 offensive and defensive evaluations each month, simulating hundreds of potential forgery attack scenarios. “Deepfake’s offense and defense is a constant process of evolution. As we advance, so does Deepfake technology. Our goal is to stay one step ahead,” explained ZOLOZ Product Director, Tao Ye.

ZOLOZ, in collaboration with the Ant Security Response Center (AntSRC), has established a reward pool exceeding one million to encourage security enthusiasts to uncover ZOLOZ Deeper’s potential vulnerabilities. Vulnerability reports can be submitted via the official website of the Ant Group Security Response Center. Tao Ye expressed, “In the digital world, security is never absolute. We aim to continuously enhance ZOLOZ’s security standards by collaborating with external security experts, thereby protecting our users’ sense of safety.”

ZOLOZ initiated its Deepfake algorithm research back in 2019 to counter simple facial attacks, such as displaying false content like photos, videos, or masks to deceive facial recognition systems. ZOLOZ Deeper is an evolution of this work, designed to counter more complex attacks. After its deployment in a major Indonesian bank, the product demonstrated its effectiveness by preventing any Deepfake risk incidents within its first month of operation.

SEE ALSO: Baidu CEO Robin Li: ERNIE Bot Users Exceed 200 Million, Releasing Three Major AI Development Tools