VK WARP (Web Application Resilient Protection)

Main article: Firewall

2024: WARP Presentation

VK has developed a system for protecting against DDoS attacks and botactivity based on artificial intelligence - WARP (Web Application Resilient Protection). The company announced this on October 30, 2024. AI models are trained on impersonal and generalized data on the behavior of users of each VK service. They distinguish malicious actions from organic bursts of activity on the site. The creation of the system is an important step in the development of artificial intelligence in VK and the integration of AI solutions to improve the sustainability of the company's services and data security. By the end of 2024, the system will be introduced into the company's projects: VK Video, OK, Zen, Mail and others.

WARP AI models analyze traffic and detect malicious requests based on constantly changing user behavior and attack patterns, and not just by the number of data packets or the speed of their arrival in a certain period of time. Neural networks allow the system to effectively and early detect malicious traffic, even if it does not differ from organic traffic.

Artificial intelligence detects and reflects attacks that bypass both predefined and scenario-specific defense rules. The technology helps to detect anomalies and fight even time-extended loads that can last up to several days instead of the previously common powerful but short-term DDoS for up to half an hour.

Neural networks recognize not only classic attacks that overload servers, channels or applications, but also attacks that combine several threats at once: parser programs that collect and systematize open data about users and thereby overload the system, brute-force programs that automatically enumerate passwords from accounts, as well as other malicious software.

VK Vice President for AI, Content and Advisory Services Anton Frolov. WARP is not just a new system for protecting against cyber attacks, but a step in creating a full-fledged AI-powered infrastructure for all VK services. Thanks to our machine learning solutions, we can quickly adapt to ever-changing cyber threats and protect the data of tens of millions of users without sacrificing platform performance.

Anton Karpov, Vice President, Director of Information Security of VK: "In smart DDoS attacks," parasitic "traffic is often indistinguishable from legitimate traffic. This is the most complex type of attack that takes into account the specifics of the service and the behavior of its users. The actions of cybercriminals not only increase the load on the selected target, but can also make it possible to enumerate passwords for accounts. We have developed a new unified system taking into account the specifics of the traffic of each of the VK projects. This significantly increases the level of protection, reduces the risk of inaccessible services for users and further increases data security. "