Infoseeds Launches AI Security Tool GOS, Supplies Major Chipmaker

Infoseeds' GOS system addresses a common challenge in cybersecurity: distinguishing genuine threats from the overwhelming volume of normal operational data. By learning an organization's typical work patterns, GOS aims to identify behaviors that deviate from the norm, rather than relying solely on static attack signatures. This approach suggests a focus on adaptive defense, which could be particularly relevant in complex enterprise environments where new threat vectors constantly emerge.

The architecture of GOS, combining graph machine learning with large language models, points to a strategy for efficiency and precision. Graph ML is used to process vast security logs and narrow down potential threats, with Infoseeds reporting that only about 0.068% of the data is then passed to an LLM for deeper contextual analysis. This two-stage process is intended to reduce the computational costs associated with LLM processing while still leveraging their capabilities for understanding the severity and context of identified anomalies.

Integration with existing SIEM and SOAR systems is a practical consideration for enterprise adoption. GOS is designed to analyze logs from a SIEM and feed confirmed threats to a SOAR system for automated response. This allows companies to augment their current security infrastructure without a full replacement, potentially easing deployment friction and preserving investments in established security operations. The reported 0% miss rate in testing, while from a company-conducted test, suggests a design goal of high accuracy.