Zscaler Report: AI-Powered Attacks Outpace VPNs as 84 Percent of Firms Shift to Zero Trust

Zscaler, together with Cybersecurity Insiders, published the ThreatLabz 2026 VPN Risk Report, based on a survey of 822 IT and security professionals. It finds that traditional VPN architecture, built on the principle of "authenticate once, trust always," fails to keep pace with attacks carried out by AI systems operating at machine speed.

The report is based on a survey of 822 IT and cybersecurity professionals and focuses on the gap between the speed of AI-powered attackers and organizations' ability to defend themselves using traditional VPNs. The report's authors note that this gap keeps widening, even though companies have spent years investing in VPN-based network security.

Attack speed outpaces patching

According to Zscaler's data, the average time to compromise a system is 29 minutes, and in the fastest documented case it took just 27 seconds. Meanwhile, 54 percent of organizations admit that rolling out a critical VPN security patch takes a week or longer, and 56 percent of respondents cite patching as their biggest operational challenge.

Only one in four companies (24 percent) has deployed AI-based monitoring, and one in five cannot distinguish an AI-assisted intrusion from an attack carried out using conventional methods at all. Just 11 percent of organizations are able to contain a compromised session to a single application, meaning that in most cases an attacker who breaches defenses can move freely across the entire corporate network.

Why employees bypass VPNs

The report also highlights the human factor. 63 percent of users admit to deliberately bypassing corporate VPN protections to reach the applications they need more quickly, creating uncontrolled risk outside the visibility of security teams. The most commonly cited reason is slow connections, cited by 29 percent of users, followed by unstable performance on mobile devices and frequent disconnections.

The report shows that 92 percent of respondents recognize the importance of adopting a Zero Trust architecture, yet it's concerning that many organizations still rely on VPNs for remote access for employees and third parties, inadvertently offering attackers an easy target - Deepen Desai, Global CISO and Head of Security Research, Zscaler

A shift toward Zero Trust

The share of companies deploying or planning to deploy a Zero Trust model rose to 84 percent, up from 78 percent two years earlier. The report's authors recommend three courses of action: limiting the blast radius of a breach through access segmentation, fully inspecting encrypted traffic instead of skipping it, and designing security so that the safe access path is also the fastest one, discouraging employees from bypassing controls.

For companies operating in Poland, the report carries practical weight, since domestic organizations still largely rely on classic VPN solutions for remote access, particularly in the public sector and among medium-sized enterprises, where migrating to a Zero Trust architecture requires both investment in new infrastructure and a shift in habits among employees accustomed to a simpler access model.

The report's findings fit into a broader trend of rising AI-related security incidents that other analytics firms are also tracking. The growing time pressure that artificial intelligence places on security teams means that classic models built around periodic patching and point-in-time access control are losing effectiveness faster than organizations can respond.

Sources: Zscaler (zscaler.com), kapitanhack.pl (kapitanhack.pl), CIO.com (cio.com)