According to international think tanks IDC and McKinsey, the market for AI-based software in cybersecurity is estimated at $93 billion, or 10% of the total volume of software solutions using AI. The share of such solutions will grow, while raising questions about the methods of working with AI, which is used in development.
Since 2015, Solar Group has been developing and integrating AI-based modules into information security solutions and services that form the architecture of comprehensive cybersecurity. More than 200 Russian companies use AI technologies as part of the vendor's software to protect their IT infrastructure. Thus, Solar Dozor has a module for recognizing graphic files: images of bank cards, passport scans, official documents with seals, medical and design documentation, which are leaders in the volume of confidential information leaks. For example, the solution processes passport data images with an accuracy of up to 99% at a speed of 135 images per second. The developers transferred the detection model to an AI server with a GPU graphics processor, comparable in power to 30 servers with CPU processors.
The R&D laboratory of Solar Group uses deep learning technologies, autoencoders for DGA domain analysis and training of large language models (LLM). Thus, LLM is used to analyze the content and categorize websites in 44 languages by 54 content categories. At the same time, the accuracy of identifying unwanted content and phishing resources reaches 95%.
To develop network security and endpoint protection products and services — NTA, EDR, XDR, and SEG — the lab develops modules based on generative adversarial networks (GAN), deep learning anomaly detection, and neural networks for graph analysis. The results are used to detect anomalies in network infrastructure, complex and targeted cyberattacks, phishing, and classify malicious processes and files.
As Maxim Buzinov, head of the R&D laboratory of Solar Group, notes, the company is developing an ecosystem of AI technologies, since data protection requirements affect most of the solutions in the product portfolio. Thus, a common data highway is formed for using AI in various products at the development and operation stage. The R&D laboratory conducts a number of studies taking into account the constantly changing landscape of cyber threats. These are developments in protecting information assets using neural networks, analyzing images, audio and video content, «smart» scanning of databases and categorization of Internet resources. Experts are developing prototypes for comprehensive detection of abnormal behavior at the level of users, network nodes, software, database operations, actions in the network infrastructure and at endpoints. Another area is the protection of user information and data from threats of compromise, biometric identification algorithms, fake detection technologies and AI assistants for decision-making in cybersecurity.
Свежие комментарии