How 100,000 Automation Servers Became a Master Key to Enterprise Data

A critical security flaw in n8n—the workflow automation darling of the AI era—has exposed an estimated 100,000 servers to complete takeover, turning what should be productivity tools into potential launchpads for enterprise-wide breaches.

The vulnerability, tracked as CVE-2026-21858 with a maximum CVSS score of 10.0, represents the second critical n8n security disclosure in as many weeks. Following Cyber Kendra's December report of an authenticated RCE flaw (CVE-2026-21877), this latest issue is arguably more dangerous—it requires zero authentication.

CVE-2026-21858 exploits a deceptively simple oversight: n8n's form and webhook handlers failed to verify content types before processing file uploads. Attackers can trick the system into reading any local file by switching a single HTTP header from "multipart/form-data" to "application/json."

What makes this particularly nasty? The attack chain progresses from reading server files to forging admin sessions—by extracting the database and encryption keys—and ultimately executing arbitrary commands. No authentication required.

"This isn't like traditional ShadowIT; ShadowAI is ShadowIT on steroids," warns Cyera Research Labs, who discovered the flaw. The researchers demonstrated how a typical enterprise knowledge-base workflow could be weaponized to exfiltrate /etc/passwd files and escalate to full code execution within minutes.

The timing couldn't be worse. With over 100 million Docker pulls and growing adoption across IT, DevOps, and marketing teams, n8n has become the invisible backbone of enterprise automation. Multiple hosting providers now offer dedicated n8n server plans, accelerating its spread into environments where security teams have zero visibility.

Here's the real danger: n8n workflows typically hold API keys for Google Drive, OpenAI, Salesforce, payment processors, and CI/CD pipelines. A single compromised instance doesn't just expose one system—it hands attackers authenticated access to everything n8n touches. One server, thousands of credentials.

Immediate Actions Required:

• Update to n8n version 1.121.0 or later immediately
• Remove public internet exposure unless absolutely critical
• Enable authentication on all Form nodes
• Audit network logs for suspicious POST requests to /webhook/ or /form/ endpoints
• Rotate all API keys and credentials stored in n8n workflows

Organizations using n8n Cloud are protected through automatic updates, but self-hosted deployments remain vulnerable until patched. Security teams should scan for TCP port 5678 and review DNS logs for n8n.cloud traffic to identify shadow deployments.

The vulnerability was responsibly disclosed in November 2025, patched by November 18, and publicly revealed on January 7, 2026—giving attackers a potential two-month window on unpatched systems.

This marks the fourth critical n8n vulnerability disclosed since late 2025. Beyond the two maximum-severity flaws (CVE-2026-21858 and CVE-2026-21877), researchers also uncovered CVE-2025-68613 and CVE-2025-68668—both scoring 9.9—highlighting systematic security challenges in rapidly evolving AI automation platforms. The pattern raises urgent questions about shadow AI deployments lurking in enterprise environments without security oversight.