China Mobile Showcases "Diting" AI Anti-Fraud System at Wuzhen Summit: Regional Fraud Cases Down 20.39% Year-on-Year

At the 2026 World Internet Conference "Light of the Internet" Expo, China Mobile's self-developed "Diting" smart anti-fraud system made a major appearance. Relying on the operator's unique advantages in communication networks and massive data resources, this AI anti-fraud platform showcased its full-chain anti-fraud capabilities spanning pre-incident warnings, precise analysis and assessment, closed-loop dissuasion, and traceback-based strikes, using digital and intelligent technology to address new types of fraud challenges and vividly demonstrating its sense of responsibility in pursuing technology for good and digital services for the people.

Today, telecom and online fraud continues to evolve and upgrade, with fraud tactics constantly donning the cloak of new technologies. New types of scams such as AI deepfake face-swapping, voiceprint synthesis, overseas phishing links, and screen-sharing-induced transfers emerge endlessly. Traditional anti-fraud models tend to focus on post-incident investigation, and often suffer from shortcomings such as delayed warnings, single-dimensional identification, and difficulty in tracing. Based on the practical needs of anti-fraud governance, China Mobile has built the "Diting" anti-fraud system, focusing on building two core capabilities: warnings for potential victims on the network side and identification of suspected fraud-related numbers on the communication side. It aims to capture risks in advance at the key links of fraud execution, conduct multi-dimensional cross-analysis through AI models, deliver high-quality technical leads for public security anti-fraud operations, and drive anti-fraud governance from "post-incident loss recovery" toward "pre-incident prevention and control and source governance."

Multiple typical real-world cases displayed at the exhibition site intuitively demonstrated the powerful operational interception capabilities of the "Diting" platform. In a fraud case in Ningbo involving someone impersonating a short-video platform customer service representative, the fraudster called citizen Ms. Chen, falsely claiming that her Douyin account had mistakenly activated a paid membership and that if it was not canceled in time, she would be continuously charged and her personal credit would be affected. Step by step, the fraudster induced the victim to disable her phone's protection, add an unfamiliar QQ contact, click an overseas phishing link, and also coaxed her into enabling screen sharing and setting up call forwarding. The fraudsters grasped her bank card information and SMS verification codes in real time and prepared to transfer more than 1.03 million yuan in deposits from the account. At the critical moment, the "Diting" platform captured the entire trajectory of high-risk behavior in real time, and its AI algorithm quickly compared it against fraud models and determined it to be a highest-level risk warning. Police-enterprise coordination immediately activated an offline emergency interception mechanism, and local anti-fraud police officers rushed to the scene and stopped the transfer on the spot, preserving all the deposits. At the same time, the platform decomposed and traced the overseas relay IP and fraud-related access paths, forming a complete evidence chain that could be used for evidence collection, achieving both loss prevention and asset protection and traceback-based fraud strikes simultaneously.

Corporate finance employee Xiao Zhang received an AI face-swapped video call from a "boss," who, citing an urgent project payment, demanded an immediate transfer of 5 million yuan. The core of this scam was AI deepfake technology—not only face-swapping, but even the tone, pauses, and catchphrases were almost flawless. At the critical moment, Zhejiang Mobile's "Diting" anti-fraud platform, through deep mining of associated information, quickly coordinated with public security authorities to implement emergency payment suspension before the user confirmed the transfer, successfully intercepting the transfer of 5 million yuan. The successful interception of one major million-yuan case after another confirms the core value of AI anti-fraud in frontline operations.

In addition to emergency dissuasion in individual cases, the "Diting" system also continues to make efforts in large-scale and routine governance, achieving two-way progress in both single-point interception and source control. On the one hand, the platform closely monitors users' fraud-related online behavior and precisely identifies potential victims who are being induced; on the other hand, it carries out full-cycle identification, labeling, and handling of fraud-related communication numbers, narrowing the communication channels available to fraudsters.

As of September 2026, in areas where the "Diting" platform supports anti-fraud governance, the number of fraud cases fell by 20.39% year on year, and case losses fell by 30.36% year on year. At present, the platform has supported 8 public security units in implementing anti-fraud operations, and has cumulatively helped successfully stop more than 15 million-yuan-level fraud cases. At the level of communication source governance, as of the end of September, the number of communication numbers involved in cases in the region has been reduced by more than 32% month on month on average, and the timeliness and precision of discovering and handling fraud-related numbers have significantly improved, cutting off the fraud transmission chain from the communication entry point.