2 min Security

Cisco aims to make AI agents safer with proposed acquisition of Astrix

Cisco aims to make AI agents safer with proposed acquisition of Astrix

Cisco appears to be looking to further strengthen its position in AI security with a potential acquisition of the Israeli startup Astrix Security. 

According to a report by The Information, cited by Reuters, Cisco is reportedly in talks regarding a deal that could be worth several hundred million dollars. With a potential valuation between $250 million and $350 million, the proposed acquisition price is significantly higher than the capital Astrix has raised to date. 

Astrix focuses on a relatively new challenge within organizations: the management and security of autonomous AI agents. These software components increasingly operate independently within IT environments and utilize a variety of tools and accounts. Astrix’s platform automatically maps these agents and analyzes how they function within a network.

In addition to inventorying, the emphasis is on identifying risks. Among other things, the software checks whether agents are unintentionally accessible via the internet or have broader permissions than necessary. In practical situations, this could mean that an AI tool that only needs read permissions can still make changes or delete files. Such anomalies are highlighted in a central dashboard, where administrators can take immediate action.

Temporary access as a security measure

An important part of the approach is limiting access over time. With so-called just-in-time policies, organizations can configure access rights to automatically expire after a short period. This reduces the risk that stolen or misused login credentials can cause long-term damage.

The platform also detects anomalous behavior, such as the mass retrieval of sensitive data. When suspicious activities are detected, the system can automatically intervene by revoking access rights. Astrix can also integrate with other security solutions, so that incident analysis and response are largely automated.

In addition, the company aims to eliminate a bottleneck in development processes. Normally, new AI agents must be pre-approved by security teams. Astrix makes it possible to define security rules in advance, allowing developers to implement agents more quickly and independently without losing control.

The potential acquisition fits into a broader strategy by Cisco to expand its AI security portfolio. Last week, the company announced the acquisition of Galileo Technologies. That company specializes in observability for AI systems and helps organizations make AI agents more reliable, secure, and transparent.