Kaspersky releases security guide for autonomous AI agents

Kaspersky has released a comprehensive guide aimed at mitigating the risks associated with autonomous artificial intelligence (AI) agents used in corporate infrastructures. The report, titled ‘AI agent security through the lens of Cyber Immunity’, was presented at the AI Everything Global event in Abu Dhabi.

Based on ‘secure-by-design’ principles, the study examines the risks inherent in increasingly autonomous agent-based systems. It analyzes real-world cyber incidents and current threat models to identify vulnerabilities such as malicious use of agents, granting excessive permissions, and accidental data deletion by AI systems.

The guide provides principles for applying cyber immunity during the design phase to establish trust within system architectures and limit the impact of potential security breaches or errors.