May 4: Zscaler, Inc., the leader in cloud security, announced participation in Anthropic’s Project Glasswing and OpenAI’s Trusted Access for Cyber (TAC) program, marking two strategic milestones that reinforce enterprise defense in the age of AI. Together, these collaborations pair frontier AI models with Zscaler’s Zero Trust Exchange™️ platform to harden software, eliminate attack surface, and defend against the next generation of AI-driven threats. The announcement comes as frontier AI models reach a point where they can find software vulnerabilities faster than humans can reading code like skilled human researchers, chaining weaknesses together, and producing working exploits in hours instead of weeks.
As part of Anthropic’s Project Glasswing coalition, Zscaler has gained access to Claude Mythos Preview, a frontier model that has already identified thousands of high-severity flaws across major operating systems and browsers. Zscaler is integrating Mythos Preview into its secure software development lifecycle to rapidly uncover vulnerabilities across its software stack and Zero Trust Exchange, further strengthening its environment and reducing customer risk. The company also intends to share findings with the broader security community and integrate Anthropic’s Opus 4.7 model into its AI Red Teaming and Agentic SecOps offerings to combat emerging AI threats with advanced AI-powered defenses.
Through OpenAI’s Trusted Access for Cyber (TAC) program, Zscaler will use GPT-5.4-Cyber within secure SDLC workflows to identify, triage, and mitigate vulnerabilities earlier and accelerate patching. Zscaler is also integrating GPT-5.4-Cyber and Codex Security into its internal multi-agent security architecture to enhance cyber defense and product hardening. Zscaler will also continue to use OpenAI’s 4.x and 5.x models across its AI Red Teaming suite including Continuous Red Teaming, Prompt Hardening, AI Asset Analysis, and the open-source Agentic Radar program and within Zscaler Red Canary’s Managed Detection and Response service, which maintains a 99.6% true-positive rate.
Combining frontier AI models with Zscaler’s Zero Trust architecture leads to better security outcomes for customers. While AI identifies and remediates software vulnerabilities, the Zscaler Zero Trust Exchange adds another layer of protection by making critical apps and software invisible to the internet with no public IP, open port, or discoverable surface for a frontier model to find. This combination provides Zscaler customers superior protection compared to obsolete VPNs and firewalls, maximizing software resiliency while systematically eliminating the internet-facing attack surface.
