As more and more enterprises begin to apply AI to core business scenarios such as customer service and operations management, enterprise AI security risks have shown a more complex and diverse development trend. In addition to traditional risks such as large model jailbreaks and sensitive information leaks, new risks such as tool abuse, intelligent agent privilege escalation, and multi-intelligent agent collaborative attacks continue to emerge.
On August 20, International Data Corporation (IDC) released the report "IDC MarketScape: Vendor Assessment of Chinese Generative AI Security Evaluation Platforms 2026". The report focuses on the Chinese generative AI security evaluation platform market and comprehensively evaluates mainstream vendors. Ant Digital successfully entered the Leader category, thanks to its technical capabilities, product maturity, and deep accumulation and large-scale implementation results in the financial, healthcare, and manufacturing industries in the field of generative AI security evaluation.
According to the information, in response to the new security risks generated by generative AI, Ant Tianjian, based on Ant's deep accumulation and practical experience in generative AI security, has built a comprehensive, trustworthy defense system covering the entire lifecycle of large models and intelligent agents, providing the industry with comprehensive and intelligent large model security testing and protection solutions.
In terms of large model security, the Ant Tianjian Large Model Security Platform has established a closed-loop system of "evaluation - defense - operation", covering the entire life cycle of large model training, inference, and deployment. It can cope with various complex attack scenarios such as large model poisoning and prompt injection attacks, while performing real-time monitoring and accurate interception of sensitive data, reducing data leakage and compliance risks. Notably, Ant Tianjian is also capable of automatically repairing model security issues and guiding large models to perform secure reasoning, fully safeguarding the security of large models.
In terms of intelligent agent security, "Ant Tianjian Loxi Guardian" relies on a lightweight plugin technology architecture to build a full-chain security monitoring and protection system for intelligent agent operating environments. It can accurately identify abnormal call behaviors, unauthorized access requests, and potential risks of malicious instruction injection, and can be seamlessly adapted to multiple types of intelligent agent operating frameworks, providing reliable support for the safe and stable operation of enterprise-level intelligent agent applications.
As of now, Ant Digital has served 100% of state-owned joint-stock banks and more than 60% of local commercial banks, among over a hundred financial institutions, and its AI security capabilities have been verified in financial-grade scenarios.
