Zscaler has officially joined Project Glasswing, a cybersecurity-focused program led by Anthropic. As a result, the company now gains access to the advanced Claude Mythos Preview model, marking a significant step toward integrating AI into modern cyber defense strategies.

With this collaboration, Zscaler plans to embed the model into its secure software development lifecycle. Consequently, it aims to proactively identify vulnerabilities across its software stack and within its Zero Trust Exchange platform. At the same time, the company will contribute its findings to other members of Project Glasswing, a group that Anthropic has carefully limited to organizations responsible for operating or protecting critical infrastructure.

Notably, this move positions Zscaler among a select group of participants leveraging Anthropic’s powerful AI model. According to the announcement, the model can identify software flaws and even generate exploit code faster than human researchers. In fact, it has already uncovered thousands of high-severity vulnerabilities across widely used operating systems and browsers.

Furthermore, Zscaler highlighted this development as part of a broader transformation in cybersecurity. The company emphasized that traditional approaches—where organizations patch vulnerabilities only after discovery—are no longer sufficient. This is especially relevant as AI systems now continuously scan internet-facing services at scale. Therefore, Zscaler argues that organizations must shift their focus toward reducing online visibility instead of relying solely on reactive measures.

In addition, the company reinforced its long-standing advocacy for zero-trust architecture. It explained that older security frameworks, which depend on firewalls, VPNs, and perimeter defenses, were designed under the assumption that attackers operate at human speed. However, Zscaler pointed out that this assumption no longer holds true in an era where AI automates reconnaissance and vulnerability detection.

As part of this architectural shift, Zscaler stressed that zero trust should function as a foundational network design rather than an added security layer. In practice, this means users do not directly connect to networks, applications remain hidden from the public internet, devices undergo strict verification, and every connection ties back to a confirmed identity.

Moreover, the company argued that eliminating public exposure—such as IP addresses, open ports, or discoverable services—significantly reduces attack surfaces. This becomes increasingly important as AI-driven tools become more efficient at identifying weaknesses in exposed systems.

Beyond internal testing, Zscaler also plans to integrate Anthropic’s Opus 4.7 model into its AI Red Teaming and Agentic SecOps solutions. These tools will help customers detect AI-driven threats, respond effectively, and manage autonomous software agents that interact with data and systems.

At the same time, Zscaler addressed growing concerns around AI-related data risks. It noted that organizations now face the possibility of their own AI tools unintentionally transferring sensitive information خارج approved environments. As a result, the company emphasized the need for comprehensive controls that inspect activity across SaaS platforms, private applications, email, and even encrypted traffic.

Additionally, Zscaler used this announcement to underline the scale of its operations. It revealed that 40% of the Global 2000 rely on its services, while its platform processes over 500 billion transactions daily. This vast data pool enables the company to distinguish between normal traffic and potential threats even before connections are established.

Ultimately, Zscaler positioned this shift as comparable to earlier disruptions caused by cloud computing, mobile adoption, and SaaS growth. However, it believes that AI represents an even faster and more transformative evolution in cybersecurity.

Recommended Cyber Technology News:

To participate in our interviews, please write to our CyberTech Media Room at info@intentamplify.com  



🔒 Login or Register to continue reading