ZOLOZ releases Deepfake defense product and sets up a "bug bounty" million-dollar reward pool

Corporate Information

Activity Area

Just released

3108

collect

Introduction: ZOLOZ teamed up with Ant Security Response Center (AntSRC) to set up a bonus pool of over one million.

On April 16, ZOLOZ Deeper, a comprehensive deepfake prevention and control product, was officially released in Beijing., in order to intercept the risk of “AI face-changing” during the user's face-swiping process, it has been the first to be applied in the field of identity security. Public information shows that ZOLOZ is the technology brand of Ant Digital, which is famous for its biometric technology and has been verified in large-scale scenarios such as Alipay App.

ZOLOZ Product Director Tao Ye

ZOLOZ Deeper has built a full-link technology system that integrates devices and clouds to deal with Deepfake security threats in systems, servers, applications and other aspects.. Through sophisticated sensor calibration and multi-dimensional feature dynamic risk control, the front-end conducts strict review of camera status, equipment and network environment, input images, etc. to ensure that the software and hardware environment is safe and risk-free, and the collected biometric data is authentic and free of tampering; During the operation, a deep learning model is used to conduct a detailed analysis of the user's facial dynamics, capture micro-expressions, muscle textures, eye movements and other subtle features, and accurately determine whether it is a real-time interaction with a real person.

In addition, Ant Group Tianji Lab will generate more than 300,000 test samples through the GAN model and hand them over to ZOLOZ Deeper for discrimination training. It will also conduct more than 20,000 offensive and defensive evaluations every month, simulating hundreds of counterfeit attack situations. “Deepfake offense and defense are a relative and constantly improving process. You are making progress, and so is Deepfake. What we have to do is to stay ahead of it.” ZOLOZ Product Director Tao Ye said.

Reporters learned at the press conference thatZOLOZ and Ant Security Response Center (AntSRC) have set up a bonus pool of over one million to support security geeks to mine ZOLOZ Deeper vulnerabilities and submit vulnerability intelligence through the official website of “Ant Group Security Response Center”. Tao Ye said: “There is no absolute security in the Internet world. We hope to unite external security technical talents to continue to improve ZOLOZ's security level and protect users' sense of security.”

According to reports, ZOLOZ started the research and development of Deepfake algorithms as early as 2019. At that time, it mainly dealt with some simple face attacks, such as presenting false content such as pictures, videos, masks, etc. in front of the camera to break the face recognition system. ZOLOZ Deeper is an iterative product that targets complex attack methods such as injection and activation. After serving a leading bank in Indonesia, the product achieved “zero leakage” of Deepfake risks within one month of its launch.

Deepfake is a combination of the words “deep learning” and “fake”. It refers to learning from a large amount of video and image data based on deep learning algorithms to fake realistic facial animations and voices. Deepfake not only increases the difficulty for the public to identify the authenticity of videos, but also further breeds the risks and hidden dangers of illegal and criminal activities such as fraud and pornography.

In recent years, global Deepfake incidents have surged tenfold. North America, Asia-Pacific and other regions have become the hardest hit areas for Deepfake attacks. The 2023 315 Party also exposed multiple “AI face-changing” fraud cases. Shi Lin, director of the Security and Metaverse Department of the Artificial Intelligence Research Center of the China Academy of Information and Communications Technology, pointed out that with the continuous optimization and innovation of deep learning algorithms, the security risks and privacy violations caused by deepfake have become more serious, and the generated content has become increasingly difficult to distinguish between real and real content. The industry and academia need to take a multi-pronged approach to continuously improve their ability to “fight AI with AI”.

Public information shows that ZOLOZ was the first company in Asia-Pacific to obtain the highest level Level 2 certification from iBeta International Biosecurity Certification Organization in 2020. In 2023, it passed the China Academy of Information and Communications Technology’s deep forgery video detection service capability inspection, meeting the requirements of safety, reliability, transparency, data protection, and clarity. Credible ability requirements in terms of responsibility, fairness, etc.