-

The Global AI Defense Challenge 2025 Launched to Fight Bad AI with Good AI

  • The competition features three tracks covering AI security technologies on images, videos, and audio, with a total prize pool exceeding $120,000 USD.

CHONGQING, China--(BUSINESS WIRE)--The Global AI Defense Challenge 2025, co-hosted by the China Society of Image and Graphics, the Cloud Security Alliance Greater China Region, and Ant Group, was launched on July 10, as part of the AI Innovation Competitions at the 2025 INCLUSION·Conference on the Bund.

With the rapid development of large models, AI technologies and applications are increasingly integrated into sectors such as financial services, healthcare, and user interactions with various gadgets and terminals. This evolution raises heightened risk of deepfake-related identity attacks, becoming a growing concern for security technology experts.

To address this global challenge and support the sustainable and reliable development of the AI industry, the Global AI Defense Challenge was hosted to facilitate collaboration between the academic and industry communities. In 2024, the competition attracted over 3,200 participants from more than 2,700 teams across 18 countries and regions worldwide.

This year's competition focuses on securing the authentication of digital identities during AI-powered interactive processes. Recognizing the evolving nature of security risks, the competition adopts closed-loop attack and defense scenarios rather than traditional static patterns. AI developers, security experts, white-hat hackers and other participants are invited to compete in the following tracks:

Track 1: Image All-Factor Interactive Authentication – Generation
Track 1: Image All-Factor Interactive Authentication – Detection
Track 2: AI Video Intelligent Interactive Authentication – Generation
Track 2: AI Video Intelligent Interactive Authentication – Detection
Track 3: Pan-terminal Intelligent Speech Interactive Authentication - Generation
Track 3: Pan-terminal Intelligent Speech Interactive Authentication – Detection

The competition is organized by Shanghai Artificial Intelligence Laboratory, Ant Security and AI Lab, Ant Consumer Finance, Tsinghua University, Shanghai Jiao Tong University, Zhejiang University, and other academic and research institutions.

Ma Huimin, co-chair of the Organizing Committee of The Global AI Defense Challenge 2025 and Vice President and Secretary General of the China Society of Image and Graphics, emphasized, “As the rapid development of AI technologies unfolds, AI forgery techniques are advancing just as swiftly, turning this into an ever-escalating 'arms race.' Therefore, enhancing AI defense capabilities and raising public awareness about AI security is urgently needed—this is the fundamental goal of hosting this competition."

Shao Xiaodong, co-chair of the Organizing Committee of The Global AI Defense Challenge 2025 and Vice President of Ant Group, added, “We have designed the competition to address scenarios where AI security on digital identities is crucial, such as in daily life services, financial services, and healthcare. Through this competition, we are committed to advancing AI security technologies and contributing to a more trustworthy AI ecosystem."

This year’s competition consists of four stages: registration and authentication, material submission, expert review, and awards ceremony. The timeline is as follows (UTC+8):

July 7–August 11, 2025: Registration and team formation
July 10–August 20, 2025: Material submission
August 20–22, 2025: Submission of model code and technical solutions by top participants
August 22–27, 2025: Results evaluation, expert review, and score verification
Early September 2025 (exact date to be announced): Competition results announcement and offline awards ceremony

Participants can register via the Alibaba Cloud Tianchi platform, the official website of the 2025 INCLUSION·Conference on the Bund, the official website of The Global AI Defense Challenge, or the official website of the China Society of Image and Graphics. The total prize pool for the competition exceeds $120,000 USD.

Contacts

Media Inquiries
Vick Li Wei
Ant Group
vick.lw@antgroup.com

Ant Group


Release Versions

Contacts

Media Inquiries
Vick Li Wei
Ant Group
vick.lw@antgroup.com

Social Media Profiles
More News From Ant Group

Robbyant and AFDE Sign MoU to Accelerate Embodied AI and Robotics Deployment Across the Middle East

SHANGHAI--(BUSINESS WIRE)--Robbyant, an embodied AI company within Ant Group, and the Arab Federation for Digital Economy (AFDE) today announced the signing of a Memorandum of Understanding (MoU) to advance the deployment, localization and commercialization of embodied AI and robotic solutions in the United Arab Emirates and the wider GCC (Gulf Cooperation Council) and Middle East markets. The agreement was signed by Dr. Ayman Ghoneim, Assistant Secretary-General of AFDE, and Yu Liang, Head of...

Ant Group’s AI Health App AQ Reaches 150 Million Users as It Expands AI-powered Day-to-day Health Use Cases

SHANGHAI--(BUSINESS WIRE)--At the 2026 INCLUSION · Conference on the Bund in Shanghai, Zhang Junjie, Vice President of Ant Group and Head of the Health Business, announced that Ant Group’s flagship AI health app AQ has reached 150 million users, handling nearly 20 million health inquiries per day. The new milestone also means AQ has grown its user base by 50% in just 8 months, since it announced 100 million users in February 2026, reflecting strong demand in China’s AI health space. Health has...

Ant Group Open-Sources Ling-3.0-flash-Fin for Real-World Financial Workflows

SHANGHAI--(BUSINESS WIRE)--Ant Group today announced the open-sourcing of Ling-3.0-flash-Fin at the 2026 Inclusion·Conference on the Bund. Grounded in actual business needs as an open model for real-world financial workflows, this release continues Ant Group's exploration of bringing AI into practical scenarios, focusing on delivering utility and high efficiency for complex tasks. Solving Real-World Challenges Through Efficiency Financial work depends on trustworthy sources, consistent definiti...
Back to Newsroom