Yiren Digital's artificial intelligence fraud detection systems intercepted over 10,300 fraudulent borrowers across 14,500 cases in 2025, helping prevent RMB165 million (approximately US$23 million) in fraud-related losses. The achievement offers a concrete window into how enterprise AI is moving beyond automation hype to deliver measurable financial protection in one of the world's most heavily regulated industries.

The company's fraud prevention framework, built on two core AI systems called Hawkeye and DiTing, represents a shift in how financial institutions approach risk management. Rather than relying solely on rule-based detection or human review, Yiren Digital has layered AI decision-making with human oversight to create what executives describe as a more adaptive and precise detection system that continuously learns from emerging fraud patterns.

What Makes This AI Fraud Detection Different From Previous Systems?

Yiren Digital's approach combines three distinct AI capabilities working in tandem. The Hawkeye system uses accumulated fraud cases and structured feedback to update screening rules and strengthen future detection. DiTing, the intelligent decision-making platform, applies AI models to analyze multidimensional information, including credit reports, user behavior, and other authorized data to support fraud identification and refined credit-risk assessment. Together, these systems process roughly 30,000 potentially risky credentials daily and identify approximately 1,500 counterfeit documents and more than 1,000 video-fraud cases each day.

What distinguishes this from earlier fraud detection tools is the scale of historical data backing the system. As of the end of 2025, the company's proprietary blacklist database contained approximately 800 million records. Hawkeye and DiTing had cumulatively identified more than 500,000 suspected fraudulent borrowers and 41,993 malicious actors associated with black-market operations. This accumulated intelligence allows the AI to recognize fraud patterns that might evade simpler detection methods.

"Risk management is one of the clearest examples of how AI can create measurable value across highly regulated financial services," said Ning Tang, Chairman and Chief Executive Officer of Yiren Digital. "Our third-generation AI fraud detection technology represents a significant advancement in financial risk management, enabling more adaptive and precise detection while continuously responding to emerging fraud patterns."

Ning Tang, Chairman and Chief Executive Officer of Yiren Digital

How Does Human Oversight Prevent AI From Making Costly Mistakes?

One critical insight from Yiren Digital's framework is that pure automation carries hidden risks in financial services. The company uses a monitoring-analysis-response-review process that combines automated identification with human review, particularly in higher-risk cases. Hawkeye analyzes fraud-risk events using historical cases, risk-assessment results, and algorithmic rules, then generates virtual work orders for fraud detection specialists to review.

This hybrid model addresses a persistent challenge in AI deployment: regulators and risk managers want to understand why a decision was made. By maintaining human touchpoints and generating reviewable decision trails, Yiren Digital creates an audit trail that satisfies both operational needs and regulatory requirements. The company's broader "All-in-AI" strategy spans pre-loan, in-loan, and post-loan processes, meaning fraud detection is embedded throughout the entire borrower lifecycle rather than applied as a single checkpoint.

Steps to Implement Enterprise AI Fraud Detection in Financial Services

• Build a Proprietary Data Foundation: Establish and continuously update a blacklist database and historical fraud case library. Yiren Digital's 800 million records and 500,000 identified fraudulent borrowers represent years of accumulated intelligence that powers pattern recognition.
• Layer Multiple AI Models for Redundancy: Use separate systems for different detection tasks, such as document verification, behavioral analysis, and identity verification. Yiren Digital's approach of combining Hawkeye and DiTing reduces the risk of a single AI system missing emerging fraud tactics.
• Integrate Human Review at Decision Points: Design workflows where AI flags suspicious cases but humans make final determinations, especially for borderline or high-value transactions. This maintains regulatory compliance and reduces liability from fully automated decisions.
• Monitor and Update Rules Continuously: Fraudsters adapt quickly, so detection rules must evolve. Yiren Digital uses structured feedback from fraud cases to update Hawkeye's screening rules regularly, ensuring the system stays ahead of new tactics.
• Measure Impact in Business Terms: Track fraud losses prevented, not just detection rates. Yiren Digital's $23 million in prevented losses in 2025 demonstrates the financial value of the investment to stakeholders and regulators.

The scale of Yiren Digital's operation underscores why financial institutions are investing heavily in AI fraud detection. Processing 30,000 potentially risky credentials daily and identifying 1,000 video-fraud cases per day would be impossible with human teams alone. Yet the company's emphasis on human oversight suggests that the future of AI in finance is not about replacing human judgment, but augmenting it with speed and pattern recognition that humans cannot match.

Yiren Digital has also made its fraud detection capabilities available as an exportable service, allowing other financial institutions and fintech companies to deploy enterprise-grade fraud protection without building the infrastructure from scratch. This move reflects a broader trend in AI adoption: companies that have invested in building robust AI systems are now offering them as services to competitors and partners, accelerating the industry-wide shift toward AI-driven risk management.

The company's underlying technology is built on its proprietary enterprise AI architecture, including the MagiCube 2.0 multi-agent platform, which provides common infrastructure for enterprise AI deployment across risk management and other core business functions. This modular approach allows proven AI capabilities to be deployed more efficiently across the organization, reducing the cost and complexity of scaling AI beyond fraud detection to other financial services functions.

Looking ahead, Yiren Digital plans to continue strengthening AI-enabled credit-risk management and governance across its credit and insurance operations, while enhancing model monitoring, explainability, and human oversight across regulated business lines. The company's focus on long-term asset quality, operational resilience, and responsible AI deployment suggests that financial institutions are moving beyond early-stage AI experiments toward mature, production-grade systems designed to operate reliably in high-stakes environments.