| Developers: | Fuzzy Logic Labs (Fuzzy Lodge Labs) |
| Last Release Date: | 2025/09/23 |
| Branches: | Financial Services, Investments and Auditing |
| Technology: | Information Security - Fraud Detection System (Fraud) |
Content |
Main article: Fraud Detection System
The Smart Fraud Detection system can detect fraudulent transactions remote banking service retail in both corporate business, mobile and SMS banking, payment terminals, processing plastic cards. It is also able to check operations through, IVR employees call center or messengers and control the actions of employees, bank including using microphones and cameras.
2025
Module "Advertising antifrod"
On September 23, 2025, Fuzzy Logic Labs (part of Rostelecom Group of Companies) introduced an additional module as part of the Smart Fraud Detection system - Advertising Antifrod. The solution allows advertisers, agencies and marketers in Russia and the CIS countries to identify and block clique fraud in real time. This reduces the financial loss of the business and increases the transparency of digital campaigns.
Advertising fraud remains a serious threat to the market: in 2024 alone, advertisers lost, according to various estimates, up to ten billion rubles due to fake user actions. This not only leads to direct costs, but also distorts analytical data, which prevents companies from making the right decisions.
Шаблон:Quote 'author=said Dmitry Reidman, director of digital business development at Rostelecom.
This module reveals the main types of unscrupulous activity: clashing, massive bot traffic, fake application installations, domain manipulation and hidden placement of ads. The system analyzes user behavior, traffic sources and technical parameters to reduce the impact of fraud on the effectiveness of campaigns.
The functionality of the solution includes comprehensive detection of advertising fraud, detection of suspicious actions in real time, automatic blocking of risk events, flexible analytics by sources and platforms, as well as convenient notifications for specialists. The solution is easy it is integrated through, API suitable for web projects mobile applications and allows you to quickly launch "Advertising Anti-Fraud" without complicated configuration.
Шаблон:Quote 'author=noted Lilia Sharovatova, CEO of Fuzzy Logic Labs.
Thanks to the ready-made connectors, the implementation of the Advertising Antifrod module takes minimal time. Data processing is carried out in accordance with the legislation of Russia and the CIS countries on personal data and modern information security practices.
Smart Fraud Detection 4.3 с ИИ
Fuzzy Lodge Labs and Rostelecom have released an up-to-date update to the flagship Smart Fraud Detection 4.3 product. This is an intelligent next-generation fraud prevention system that makes it possible to further analyze user behavior to more confidently identify fraudulent transactions. Rostelecom announced this on August 11, 2025.
The Smart Fraud Detection (SFD) 4.3 system version uses neural networks and artificial intelligence technologies. They can detect fraud according to known patterns, as well as independently create new rules, identify hidden behavioral models and anomalies. This approach significantly improves the effectiveness of protection, making the fight against frode more dynamic and reliable.
{{quote 'author=said Dmitry Reidman, Director of Digital Business Development at PJSC Rostelecom. | Digital threats are constantly becoming more complex and evolving. We try to anticipate these threats by turning technologies such as artificial intelligence and machine learning into a powerful defense tool. Our anti-fraud solution is the result of many years of experience and a deep understanding of the mechanisms of fraud. Continuous improvement allows us to provide businesses with reliable security and confident growth in the era of digital challenges,}}
Fuzzy Logic Labs uses machine learning (ML) algorithms in SFD to build models of customer behavior and detect anomalies. The system is constantly learning from new data, which ensures its adaptability and reliability. For example, paying attention to a suspicious accumulation of small transfers to a limited number of recipients in a short period of time.
To work with modern data and detect fraud, the ML model SFD uses more than 250 internal profiles and features that help accurately assess the risk of each operation. Among them are key groups: client, sender and recipient (determined by details), merchant, device, IP address and internal communications.
Another important component in creating an antifrod system is graph neural networks. They allow you to recognize networks of related fraudsters and build relationships between transactions, devices and users. This helps identify complex fraud schemes that can be invisible when analyzing individual transactions.
Inside the SFD system, two methods are used to detect fraud: Bayesian tree and gradient boosting. The system analyzes the data and, with the help of decisive trees and logistic regression, determines how likely fraud is on a case-by-case basis. This helps to accurately identify suspicious transactions and take measures to prevent them.
Шаблон:Quote 'author=noted Lilia Sharovatova, CEO of Fuzzy Lodge Labs.
2024: Smart Fraud Detection 4.2 with the ability to re-check operations according to 369-FZ
Rostelecom and Fuzzy Lodge Labs have updated the Smart Fraud Detection fraud prevention system to version 4.2. Rostelecom announced this on July 24, 2024.
The updated version not only reveals the latest trends, financial attacks but also provides advanced analytics of customer behavior, providing even more reliable protection for business and user interests. Smart Fraud Detection 4.2. fully complies with current legal requirements.
Version 4.2. received an important function: now the system can re-check operations according to the 369-FZ. This allows banks and other financial institutions to respond quickly to suspicious transactions, suspending them for additional verification in case of suspicion of fraud or violation of legislation. This approach prevents possible losses and protects the interests of customers.
The updated version of Smart Fraud Detection fully complies with the requirements of the Central Bank of the Russian Federation. If the operation was paused by the system but is later confirmed by the client, it can be automatically or manually sent for re-validation. Based on the results of the audit, a final decision will be made to conduct such an operation.
Smart Fraud Detection has gone beyond the banking sector and is already widely used in various areas, including classic retail chains and modern marketplaces. In the updated version of the system, it became possible to analyze the customer's purchasing behavior based on the product items in the order or check. In this version of Smart Fraud Detection, the data type List was implemented.
The flexibility of the List allows you to adapt Smart Fraud Detection to different business tasks that require in-depth analysis of structured data, for example, to evaluate a loan application, borrower guarantors or its loan security.
Шаблон:Quote 'author=said Lilia Sharovatova, CEO of Fuzzy Lodge Labs.
Version 4.2 of the Smart Fraud Detection system also received additional pre-installed reports: distribution of risk assessments and distribution of transactions with resolutions. They will allow users to more flexibly manage risks, reduce the number of false positives (False/Positive) by applying the SFD risk assessment. The functionality of managing rules and lists, a key element of SFD, has also been improved. Users can now track the number of positives over a specific period, verify that conditions and actions are met correctly, and notify administrators of any problems or errors.
{{quote 'author=noted Sergey Parfyonov, Technical Director of Fuzzy Logjik Labs. | This system update has become more fault-tolerant and secure. This once again shows that our developers are constantly working to improve the product, as well as to pay attention to the needs of users. I am confident that this update will make working with the Smart Fraud Detection system even more convenient and effective,}}
Smart Fraud Detection 4.2. - the result of the continuous work of the teams "Fuzzy Lodge Labs" and "Rostelecom" to improve the tools for protecting business from fraud. With each update, companies strive to provide customers with even more reliable and effective solutions that can withstand the most sophisticated deception schemes.
2022
Anti-fraud systems in the fight against credit fraud
According to data from open sources, in 2021 it was possible to prevent the theft of 24.2 billion rubles of customer credit funds. Sergey Parfyonov, Technical Director at Fuzzy Logjik Labs, spoke about the new developments of Fuzzy Logjik Labs in the field of protection against credit fraud using the Smart Fraud Detection anti-fraud system. Read more here.
Red OS Compatibility
The Russian IT-Companies Fuzzy Lodge Labs"" and RED SOFT"" confirmed the correct operation of the countermeasures system to fraud in the banking channels of Smart Fraud Detection on. operating system RED OS This was announced by RED SOFT on April 12, 2022. More here
2021
Add Loyalty Protection
On April 12, 2021, Fuzzy Lodge Labs introduced another function of the Smart Fraud Detection anti-fraud system to combat fraud in bonus systems and loyalty cards. The solution is based on a combination of rule method, machine learning, and working with object profiles.
Rules are configured to label suspicious activity or known attack patterns using the parameters of specific client and/or employee actions and the analysis of dynamically calculated objects. Machine learning methods allow you to detect anomalies in the behavior of customers and employees of the organization, without requiring long-term configuration and support, automatically adapt to changing patterns of attacks by attackers.
Working with dynamic profiles includes storage objects to describe an unlimited number of elements and arrays of maximum quantity/specified data depth. This allows
- build object profiles: user, user device, map, event geolocation, employee, store, purchase type, etc.;
- monitor typical and atypical parameters, the most important and frequent interactions between objects;
- profile on the basis of operations with "points movement" and other events (for example, changing personal data, registration of a mobile application).
Suspicious/Froda Activity Rates
- Abnormal number of transactions during the period of card points accrual, presence of several simultaneously working cards for the client
- Customers with an atypically large number and amount of purchases, an atypical number of returns using a card,
- Information about the device and geo-positions when using mobile and WEB web applications
- Frequent use of the same card at different outlets where points were awarded over time
- Atypically high percentage of loyalty card usage at the outlet
- Card Balance Check Operations when Teller Transfers a Shift (Known Fraud Preparation Indicator)
- Purchase of goods that are not similar in properties (for example, they take all types of fuel at gas stations on the same card)
- Write off bonuses as soon as the activation period expires
To protect loyalty programs of large retailers, banks, developers, Fuzzy Lodge Labs uses proven technologies to control external and internal fraud in the financial sector. Attacks on loyalty programs are no less diverse and sophisticated than fraud in banking systems. They involve not only external "actors," but also, in many cases, the employees of the organization themselves.
Smart Fraud Detection 3.5 with dynamic calculation objects
On April 8, 2021, Fuzzy Lodge Labs announced the update of the Smart Fraud Detection system to version 3.5. Updates include additional transaction parameters for the NWS, features for user convenience, and technical developments in the rules module and behavioral profile assessment.
- To comply with the NSPK requirements for the Fast Payments System, the transaction parameters for transferring the amount and currency of the commission have been added to the Smart Fraud Detection system. Parameters are now available in the interface for use in rules and other calculated parameters.
- For the convenience of users, the graphical forms of the interface have been redesigned, note templates have been added for use in incidents, rules and lists, a reference book has been developed to add/remove attributes in lists in the incident card.
At the request of users, an additional functionality has been developed for the automatic formation of existing and new reporting forms (Scheduled Reports), followed by sending them by email or saving them on a network resource.
The following reporting forms have been added to the system:
• Monitoring Night Tasks - Statistics data on Night Tasks and Data Calibration
• Transaction Processing Time - Transaction Processing Time Data
- Version 3.5 of Smart Fraud Detection introduces Dynamic Calculation Objects. This feature allows you to work with your own custom storage objects. Dynamic objects are used when calculating additional parameters according to their own algorithms for detailed analysis of behavioral profiles.
For the Rule Generator function, the system adds the ability to create new generation requests by copying existing ones.
Function for evaluating behavioral profiles
On March 10, 2021, Fuzzy Lodge Labs introduced a function for assessing behavioral profiles - Dynamic Calculation Objects.
This feature allows you to customize your own storage objects for calculating behavior data and is generally designed to collect long-term information on:
- simple objects (one transaction attribute) "client," "recipient," "terminal," "country," "phone," "IP..."
- composite objects (combination of several transaction attributes): "card + terminal," "amount + BIC," "city + recipient."
So, for example, the following information will be available on the composite object "map - terminal": the time of the first operation, the time of the last operation, the volume of operations per day, per month, the ratio of the volume of operation per month and per day, etc. This allows you to highlight atypical long-term actions and track anomalies in behavior, noted in "Fuzzy Lodge Labs."
The storage object can be described by an unlimited number of elements and includes arrays of the maximum amount of data or a given depth.
You can work with storage objects - add, update, fill - online through the system interface, storing in inmemory. So within a few minutes you can create new calculation algorithms for work.
Adding 3D Secure 2.0
On March 9, 2021, Fuzzy Lodge Labs announced that it had supplemented the monitoring fraud system with data analysis of 3D Secure 2.0.
Smart Fraud Detection now uses 3D Secure 2.0 data to further monitor transactions and assess the risk of payments made when making online purchases (Risk-Based-Authentication).
This system function allows
- In real time, create more accurate profiles of objects in various channels, as well as connections between them (customers, recipients of payments, merchants, IP addresses, details and geolocation of client devices...)
- Define cross-channel recurring payments, typical devices, IP movement speed
- Check payments based on details and devices of known fraudulent transactions carried out in other bank channels
- Rank transactions from least risk to most risk by assigning a risk score to each event
With this extra layer of online fraud protection:
- A payment transaction can be performed with or without the least risky verification method
- Recurring payments are made without the need for a one-time password or other authentication tool
- Banks will be able to significantly save on SMS mailings
2020
Function for monitoring employees at a remote location
Fuzzy Lodge Labs Company on October 14, 2020 introduced the Smart Fraud Detection system function to control employees at remote work.
Smart Fraud Detection monitors employee behavior and detects atypical actions when performing financial transactions using cameras and microphones. Photo, video and audio data are processed, anomalies are detected, incidents are created for further investigation and setting up rules.
According to the photo and video data, the system authenticates the employee and the client, analyzes the emotional state, determines the classes of objects in the frame. For example, the system determines the photographing of the screen by the dynamics of the employee's pose and the detection of the phone or camera. According to audio data, the system determines the type of room, the number of speaking people, the nature of activity at the computer (according to the patterns of work on the keyboard/mouse). The system recognizes the individual frequency of the employee's voice and detects anomalies, for example, increased emotionality or an atypical sound background for a particular employee.
The system creates a profile for each employee with the department, store, time zone, and working time. If an anomaly is detected, the system displays a centralized event history for the selected employee. So, for each employee, work during working and non-working hours, the presence and type of violations are monitored.
Fraud Tracking Feature
Fuzzy Lodge Labs Company on October 14, 2020 introduced a function to track fraudulent transactions with the Smart Fraud Detection system - Rule Generator. In contrast to manually writing rules to identify fraud, the rule generator automatically generates hypothesis rules and sequentially optimizes them, improving them after each iteration. If, for example, the optimization of social engineering rules is carried out, then the system, based on the results of the work, will propose other rules for better finding such operations with a minimum false response.
The rule generator uses two types of rule checks: Antifraud and AML. The search takes place according to the specified transaction types, service channels and departments of the company. Rules are generated/optimized off-line and are convenient for further analysis. For example, queries can be created by copying and saving a dataset for re-runs; You can expand the list of available operators, export results, and view general transaction statistics.
Notes
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |