Ant Group’s Security Parallel Aspect Fusion AI: A Case Study Selected at the 2024 World Intelligent Industry Expo

On June 20, 2024, the World Intelligent Industry Expo opened in Tianjin and announced the “Find Intelligent Technology Innovation Application Outstanding Cases” award, with Ant Group’s “Security Parallel Aspect Fusion AI for Threat Detection” being selected as a typical case.

The award‑selection activity, jointly organized by the Expo committee, the China New‑Generation Artificial Intelligence Strategic Development Research Institute, the China Software Industry Association, and the China Cyberspace Security Association, invited global submissions on smart manufacturing, intelligent connected vehicles, and intelligent cybersecurity, receiving 990 cases evaluated by over 40 renowned academicians and experts.

In recent years, network threat risk has become increasingly severe and complex, with traditional detection methods suffering from high false‑positive rates, weak unknown‑threat discovery, and poor explainability due to insufficient internal visibility and limited intelligence.

Since 2019, Ant Group pioneered a security parallel aspect system and later deeply integrated artificial‑intelligence techniques, establishing a multi‑layered network security depth‑protection architecture centered on “security parallel aspect fusion intelligence.”

The solution first captures multi‑dimensional data from systems, networks, and applications via aspect technology, extracts network entity credentials, and constructs complete attack chains through correlation analysis.

Next, a security large model fuses expert intelligence with machine intelligence, leveraging the strengths of algorithms and domain experts to improve the detection of unknown threats.

Simultaneously, an application‑centric security baseline is built; the large model uses this baseline and expert knowledge to identify normal business behavior, thereby reducing false alarms.

Finally, system, network, application, and intelligence data are used to build a security adversarial knowledge graph, enhancing model explainability and enabling the discovery and tracing of attack patterns.

In practice, facing challenges such as real‑time detection of trillion‑scale data, rapid response to strong adversarial attacks, financial‑grade stability, and quantifying intrusion detection levels, the “Security Parallel Aspect Fusion Intelligence” framework has consistently delivered outstanding results, achieving a risk perception rate for anomalous behavior far exceeding traditional techniques and an automated analysis rate above 95%.