Kaspersky AI agent cybersecurity developments

Kaspersky has addressed the security implications of autonomous artificial intelligence (AI) agents in corporate environments. The company released a guide titled ‘AI agent security through the lens of Cyber Immunity’ at the AI Everything Global event in Abu Dhabi. This report focuses on ‘secure-by-design’ principles to mitigate risks such as malicious agent use, excessive permissions, and accidental data deletion.

Building on these security concerns, Kaspersky experts have predicted a shift toward an era of ‘AI vs. AI cyber warfare’. Sergey Lozhkin, Head of Global Research and Analysis for Kaspersky's APAC and META regions, suggests that AI agents may manage the entire lifecycle of cyberattacks—including intelligence gathering, social engineering, system infiltration, and data theft—while simultaneously being deployed as advanced defense mechanisms. This evolution suggests a future of highly automated combat between AI-driven attack and defense systems with minimal human intervention.

In response to the rapid adoption of AI, industry players are introducing tools to manage risks associated with AI agents and ‘Shadow AI’. Microsoft is utilizing its security ecosystem, including Microsoft Purview and Microsoft Defender, to help enterprises govern AI workloads and monitor agents with excessive permissions. Concurrently, Kaspersky has updated its products to counter AI-driven threats. By adding profiles to the Kaspersky Threat Intelligence Portal, the company aims to describe how attackers exploit these tools, while its Endpoint Detection and Response solutions are designed to detect AI agent activity and identify dangerous behaviors.