Rubrik Launches Code Guardian to Counter AI Security Threats

Rubrik introduces Code Guardian under Project Hourglass to scan AI-generated code for security flaws and help enterprises prepare for advanced AI-driven attacks.

By Divine Media 24 ·October 10, 2026 ·2 min read ·138 views ·Updated 27 October 2026, 6:18am

Cybersecurity company Rubrik has introduced Rubrik Code Guardian, an AI security tool designed to help enterprises defend against advanced threats associated with increasingly capable AI systems. Announced on October 8, 2026, the product expands the company’s cybersecurity initiative, Project Hourglass.
Code Guardian is designed to scan AI-generated code for security vulnerabilities within isolated, air-gapped environments. These environments restrict external network access, helping organisations analyse potentially risky code while reducing exposure to connected systems.
According to Rubrik, the tool aims to help businesses prepare for sophisticated attack chains involving AI agents. It also offers codebase recovery capabilities, allowing organisations to restore code after security incidents or other disruptions.
The launch comes as businesses increasingly use AI tools to generate and modify software. While AI-assisted development can speed up coding, it can also introduce vulnerabilities that may be difficult to identify without dedicated security checks. Automated scanning and controlled testing environments can help security teams detect issues before vulnerable code is deployed.
Code Guardian forms part of Rubrik’s broader Project Hourglass programme and comes with access to Rubrik Agent Cloud. The company is also a participant in Anthropic’s Project Glasswing initiative, which focuses on addressing cybersecurity challenges associated with advanced AI capabilities.
The term “Mythos-level threats” refers to the advanced risks associated with highly capable AI systems that may be used to discover or exploit software weaknesses. Security tools designed for these risks are intended to help organisations strengthen their defences as AI capabilities develop.
However, the effectiveness of any security tool depends on its detection performance, integration with existing systems and ability to respond to evolving attack methods. Enterprises will need to evaluate these factors as they consider new AI-focused cybersecurity solutions.