Share with your CISO
Cloudbrink is betting that existing ZTNA and SASE architectures (zero trust network access and secure access service edge, the two dominant frameworks for controlling who gets into what on a corporate network) are structurally blind to the AI agent layer now running inside enterprises. Its answer is Veraify, an endpoint-first AI security platform that discovers sanctioned and shadow AI tools, enforces data loss prevention across prompts and uploaded files, and applies identity and device policies through a single control plane. Cogent Security’s Field CISO is named as an early adopter. Current Cloudbrink customers get a free trial for a limited time.
What this means for your business
The 69% figure Veraify cites from Gartner, that roughly seven in ten organizations suspect or have confirmed employees using prohibited generative AI tools, is the sharpest thing in this announcement, and it’s the number that decides whether this story is about you. If your organization has deployed Microsoft Copilot, GitHub Copilot, or any coding assistant that runs as a desktop process rather than through a managed browser, your existing SASE stack almost certainly has no visibility into what those agents are doing with corporate data. That’s the exposure Veraify is targeting, and it’s real regardless of whether Veraify is the right fix.
The architectural argument here is worth taking seriously, even accounting for the fact that Cloudbrink is selling into the future it describes. Browser-based and cloud proxy approaches intercept traffic at the network edge, but AI agents increasingly operate at the endpoint level, calling APIs, reading local files, and writing to data stores without ever touching a web browser. That’s not a gap that a firewall rule closes. An endpoint agent approach, where policy enforcement lives on the device itself, is a defensible response to that pattern. The open question is whether Veraify’s “top 100 AI catalog” plus automated discovery is actually comprehensive enough to catch the long tail of embedded AI capabilities showing up inside line-of-business applications.
The vendor category Veraify is entering is genuinely crowded, with players like Nightfall, Cyberhaven, and Prompt Security all making adjacent claims. What distinguishes Veraify on paper is the combination of Cloudbrink’s high-performance connectivity layer with endpoint-level AI visibility, a pairing that matters most for organizations where latency-sensitive workloads have already made legacy SASE proxies painful. If your SASE renewal is coming up in the next 12 months and AI agent governance is an open gap on your risk register, that’s the budget conversation this announcement is trying to interrupt, and the free trial for existing Cloudbrink customers is designed precisely to get a foot in the door before that decision closes.
Concept deep-dive: Shadow AI
Shadow AI is the enterprise AI equivalent of shadow IT, employees using AI tools that IT hasn’t approved, often because the approved options feel slower or less capable. Unlike shadow IT (an unauthorized SaaS subscription), shadow AI carries a distinct data risk because the exposure happens inside the prompt itself. Typing customer records or proprietary code into an unmanaged AI tool sends that data to a third-party model. Governance frameworks that only inspect network traffic after the fact arrive too late to prevent that transfer.
Based on reporting from Veraify Launches AI-Native Security Platform Powered By Cloudbrink, originally published 2026-07-23 10:27:00.
