聚焦 AI 时代大规模协作难题，蚂蚁数科全年超 20 篇技术论文入选国际顶会

Core Insights - Ant Group's research division has gained international academic recognition, with 22 papers published in top-tier journals and conferences since 2025, over 85% of which are classified as CCF A-level, focusing on key challenges in large-scale intelligent collaboration in the AI era [1][2] Group 1: Research Achievements - Ant Group's research addresses three major challenges in the AI era: establishing trust in distributed environments, responding to new security threats posed by AI technology, and enhancing the efficiency of large-scale collaborative decision-making [1] - The research results this year focus on blockchain consensus protocols, AI security, and trusted computing, providing new solutions for large-scale intelligent collaboration [1] Group 2: Blockchain Consensus Protocol - In collaboration with Tsinghua University, Ant Group developed the Rondo protocol, which provides secure, fair, and unpredictable "random numbers" for distributed networks with minimal resource consumption, laying the foundation for efficient collaboration among agents without a trusted third party [1] Group 3: Security Frameworks - To address security risks from deep fakes, Ant Group has designed a new learning framework that decouples features such as texture, color, and edges in image information, significantly increasing the difficulty of solving deep fake issues [2] - Ant Group introduced a "key forgetting" mechanism that allows detection models to ignore common forgery traces and focus on deeper forgery features, enhancing the ability to identify and warn against unknown forgery techniques [2] Group 4: Zero-Knowledge Proofs - In collaboration with East China Normal University, Ant Group achieved two key breakthroughs in the field of zero-knowledge proofs, creating the industry's first verifiable graph data query algorithm and the first verifiable federated learning framework (ZKSL), establishing a trusted protocol layer for inter-organizational agent collaboration [2] - These advancements enable rapid generation of verification proofs, ensuring the trustworthiness and efficiency of collaborative decision-making [2] Group 5: Collaborative Mechanisms - Ant Group has established joint laboratories and targeted research projects with universities such as Tsinghua University, Zhejiang University, and East China Normal University, promoting cutting-edge technology exploration and implementation through industry-academia collaboration [2]