Artificial intelligence is already involved in one in six hacks worldwide — and the trend is accelerating

Hackers are increasingly using artificial intelligence systems when attacking corporate networks; the report says bots and agents have nearly doubled in activity scanning protective perimeters for vulnerabilities.”KommersantAt the same time, artificial intelligence is also actively used in network defense tools – it can help speed up intrusion detection and reduce the burden on traditional monitoring tools.

Photo credit: Milad Fakurian / unsplash.com

Traffic generated by AI bots and AI agents acting as intelligence agents increased by 187% in one year; InfoWatch analysts estimate that traffic by autonomous AI agents scanning security perimeters increased nearly 80 times. If by 2023 the share of malicious bots in traffic to protected resources of Russian companies was 30%, by 2025 this proportion will reach 41%. In today’s world, approximately one in six recorded system incidents and hacking attacks are carried out with the direct involvement of artificial intelligence tools; in 86% of cases, phishing attacks designed to initially penetrate a network were prepared or optimized using AI.

Artificial intelligence is 70% more efficient than humans in identifying zero-day vulnerabilities. The artificial intelligence performing the attack disguises its traffic as legitimate, making intrusion detection difficult. InfoWatch said that the tactical advantage of this “cyber arms race” still lies in defense: the use of advanced threat detection tools (NDR/XDR) doubles the speed of detecting anomalies and targeted attacks, artificial intelligence filters on emails help block up to 94% of phishing, the risk of insider threats caused by behavioral systems is reduced by 45%, and the workload of monitoring center analysts is reduced by 60%.

Experts say the effectiveness of AI-based defenses depends on the stage at which attacks are stopped – the earlier the better. Russian companies use artificial intelligence in behavioral analysis, anti-fraud and email traffic analysis. It is important to distinguish between AI systems that provide recommendations and AI systems that take direct action – if the system gains access to company data or APIs, then the security aspect becomes a mandatory part of the project cost.