Kaspersky adds AI agent detection and control to its endpoint and security products
Original title: Deploying AI agents securely: how Kaspersky products mitigate AI threats
Kaspersky says over 2500 AI-related vulnerabilities have been published since 2025 and its products can now block or monitor agents like Claude and Codex.
Kaspersky describes new capabilities across its products for detecting, tracking and restricting AI agents in corporate environments. Application Control and Web Control in Kaspersky Endpoint Security (also sold as Kaspersky Next) can block AI agent launches and web access per employee group, while EDR rules flag suspicious agent behavior. The company also added AI agent profiles to its Threat Intelligence Portal and says Kaspersky Scan Engine analyzes installed apps and LLM files for malicious code. Kaspersky telemetry places Claude, Claude Code and Codex at the top of its September 2026 AI agent popularity ranking.
Unconfirmed: The vulnerability count and popularity rankings come from Kaspersky's own telemetry and have not been independently verified.
- Kaspersky telemetry ranks Claude, Claude Code and Codex as the most popular AI agents.
- More than 2500 vulnerabilities in LLMs, AI tools and plugins were published since 2025, Kaspersky says.
- Application Control and Web Control can block or allow AI agents per employee group.
- Kaspersky Scan Engine checks applications and LLM files for malicious code.
- Threat Intelligence Portal gained profiles describing AI agents and attacker abuse.
- 1.Use Application Control KL categories to allow or block AI agents for specific employee groups.
- 2.Scan AI agent skills and LLM files for malicious instructions with Kaspersky Scan Engine.
- 3.Track AI agent launches and activity with endpoint detection and response.
- Vendors
- Kaspersky
From the original article
11 min read at securelist.com
We look at best practices for deploying AI tools securely in corporate environments, with Kaspersky Security Center and Kaspersky Endpoint Security as examples. We also cover ways that EDR and TIP can detect AI agents.
securelist.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.