Artificial intelligence "Yandex Market" recognizes suspicious orders in PVZ

2025

Using AI to identify suspicious orders issued with postpayment

Yandex Market began using artificial intelligence to identify suspicious orders issued with postpayment. The decision will prevent fraudulent schemes when goods are ordered at the point of issue (PVZ) with payment upon receipt, stolen, and the owner of the point and the seller incur losses. The company announced this on December 25, 2025.

To do this, the marketplace trained the ML model on data on previous purchases and anomalies in buyer behavior. The system analyzes each new order in the PVZ and assigns it a risk level. If the operation looks suspicious, it automatically notifies the QA specialist. The employee examines the case and makes the final decision whether it is possible to place an order and what payment method is available for him.

The task of this system is to promptly respond to various frauds with orders, including from users who are trying to fraudulently pick up expensive goods from PVZ without payment.

Artificial intelligence on Yandex Market also helps in the fight against counterfeit. To do this, the site, for example, uses an ML system to automatically process and classify user complaints about non-original goods. The joint work of algorithms and specialists brings results: the share of complaints about counterfeit is decreasing and as of December 2025, the goad does not exceed 0.08% of all orders.

Implementation of ML to automatically process and classify buyer complaints about potential counterfeiting

Yandex.Market has introduced a machine learning system (ML) to automatically process and classify buyer complaints about potential counterfeiting. The technology allows employees to quickly respond to appeals and, if necessary, hide goods from the site, which reduces the chance of purchasing a non-original product for the buyer and makes shopping on the marketplace safer. The company announced this on November 27, 2025.

Thanks to comprehensive measures, including the work of algorithms, the share of complaints about counterfeit products from the total number of orders of the Market continues to decline and as of November 2025 is only 0.08%.

AI accurately determines the essence of the buyer's complaint, even if it is difficult or incorrectly formulated. Then the algorithms - if there are other requests - form a detailed report on the store and transmit the information to the quality control specialist. The employee is left to double-check the data and make a final decision on measures against the seller, up to his blocking. Previously, the initial processing of complaints took place completely manually.

The system takes over the routine and helps us quickly see the full picture of the seller and incoming requests. At the same time, the technology sorts 2.3 times more messages and provides collected information for decision-making. In addition, this reduced the financial costs of their processing by 1.5 times, - noted in Yandex Market.

Running an algorithm to protect against scammers using AI

Yandex.Market has developed an algorithm based on Yandex technologies that can track frauds even before they bring any harm to buyers, partners or the marketplace itself. Yandex.Market announced this on May 14, 2025.

The algorithm analyzes the data received throughout the product path from warehouse management systems, logistics, orders and others. Based on this data, it calculates potential scenarios for goods movement. If they detect a deviation from the norm, for example, a delay in delivery or a change of route, the algorithm signals this to the security service. She, in turn, immediately blocks transactions with the product and initiates an investigation. Despite the fact that the algorithm was only recently taken out of testing, it has already prevented a real case of fraud.

The described algorithm is only part of a large system that ensures the safety of buyers and partners of the marketplace. The company conducts dozens of internal and external security audits each year, identifying and addressing potential vulnerabilities in its IT systems and logistics infrastructure. In parallel, work with partners is underway - for example, a free security training course is available for owners and control points of issue of orders (PVZ). It helps to minimize the risks of crisis situations and increase the overall level of protection and efficiency of the PVZ.