Zscaler Expands Project Glasswing, Integrating Claude Mythos Preview into Secure Development Cycle
Zscaler has announced its integration of Claude Mythos Preview into its secure development cycle, a strategic move highlighting a radical architectural conclusion: in a rapidly evolving landscape where vulnerabilities emerge at machine speed, traditional fix-after-the-fact security cannot keep pace. The structural response? Minimize the attack surface to the extent that adversarial AI has nothing to exploit.
By incorporating Mythos Preview into its software development lifecycle, Zscaler aims to identify vulnerabilities within its own application stack and its Zero Trust Exchange platform before sharing findings with the program community. This initiative reinforces a belief Zscaler has championed for years, which the advent of Mythos now necessitates. “When the cloud era began, the industry insisted that the traditional perimeter would hold; it didn’t. When mobile and SaaS flourished, the industry claimed that VPNs would adapt; they failed to do so. Every twenty to thirty years, architecture must evolve, and those companies that adapt end up dominating the next decade,” noted Zscaler in its launch blog. The operational integration into Project Glasswing is a practical realization of this perspective.
The company relies on a substantial detection infrastructure, processing over 500 billion transactions daily, with 40% of Global 2000 companies among its clientele. This immense volume of signals serves as the foundation for its behavioral detection capabilities. Zscaler intends to leverage this data to develop a response to threats exacerbated by AI, making its architecture fundamentally less vulnerable rather than merely hastening the patching cycle.
Mythos Preview’s Role in Zscaler’s Development Cycle
Zscaler’s engagement with Project Glasswing encompasses two main scopes. The first is defensive, focusing on the publisher’s own assets. Mythos Preview is deployed early in the development cycle to identify vulnerabilities within the Zero Trust Exchange code before production. The results are shared within the program’s community, following the collective sharing principle that underpins Project Glasswing. The second scope is commercial; Zscaler also incorporates Claude Opus 4.7, the new production model from Anthropic, into its AI Red Teaming and Agentic SecOps offerings. These products are designed to detect and mitigate threats posed by autonomous AI agents capable of accessing data and interacting with other systems without direct human oversight.
This dual deployment exemplifies the strategy Anthropic has devised around the Mythos family. Mythos Preview remains under restricted access for high-sensitivity offensive-defensive operations, while Opus 4.7, equipped with enhanced cyber safeguards and a deliberately reduced offensive capacity compared to Mythos, is positioned for legitimate SecOps applications. For Chief Information Security Officers (CISOs) evaluating security automation tools, this distinction between the two models is not merely technical; it delineates what can be deployed in production within a Security Operations Center (SOC) and what remains within the realm of controlled research.
A Two-Tier Cyber Defense
The gradual expansion of Project Glasswing creates a defensive asymmetry disadvantageous to organizations outside the coalition—a form of two-tier cyber defense. Members proactively detect and remediate critical vulnerabilities within their assets before these flaws become known to the broader market. For European firms utilizing a Zero Trust Exchange platform audited by Mythos, this delta results in a higher level of fortification compared to non-member platforms. Currently, no equivalent European initiative offers similar access to critical infrastructure operators on the continent.
However, the implications of the Cloud Act loom just as significantly for coalition members. Integrating Mythos Preview or Opus 4.7 into a secure development cycle or SOC entrusts Anthropic with data concerning vulnerabilities in one’s infrastructure. As a U.S.-based entity subject to the Cloud Act, this reality necessitates a specific risk analysis prior to deployment for vital European operators and entities under NIS2 obligations. The regulatory exposure extends not only to data processed in production but also to the sensitive vulnerability information itself. The defensive capabilities provided by Glasswing and the legal sovereignty constraints present two sides of the same coin, and European CISOs will need to address them simultaneously.



