CVE-2026-74014

The CVSS 9.9 for CVE-2026-74014 is deserved, but it masks a deeper failure: arbitrary file upload from subscriber accounts shouldn't exist in this class of application at all. IT Residence is a ticketing system where subscribers are end-users filing requests — the requirement to attach screenshots got implemented without any of the controls that would make it survivable. No extension whitelist. No transcode to strip payloads. Storage presumably within webroot. The developers thought in terms of what users needed to attach, not what an attacker could do with the endpoint. This is the specification genealogy problem in action: a business requirement enters the backlog as 'attach screenshots to tickets,' gets handed to engineering as a user story, and never carries its threat model with it. The fix isn't adding a security review at the end — the requirement itself should have specified a sanitized pipeline with extension validation, webroot isolation, and CSP headers. That context lives in a different document that never gets written. But the vulnerability isn't just the missing controls — it's the subscriber role itself. These accounts were provisioned when the product deployed, probably by a different team in a different security posture. As the application grew around them — SSO got wired, LDAP connected, new integrations added — the subscriber endpoint absorbed all of it silently. No one audited what a 2026 subscriber account can actually reach because nobody thinks of subscribers as a threat surface worth revisiting. That's where the real exposure compounds: the blast radius from a subscriber compromise extends into everything that ticket system touches — credential reset links in history, vendor contacts, internal hostnames, API keys for integrated tools. The low EPSS score likely reflects a niche product with small deployment footprint, not reduced severity. One exploitable instance is sufficient for a critical rating regardless of total deployments. And once inside, the attacker lands in a connected environment where post-exploitation lateral movement is the real risk, not the upload itself. The persistence of CWE-434 (Unrestricted Upload of File with Dangerous Type) across two decades isn't because teams don't know file upload is dangerous — it's because the knowledge that 'this is documented in OWASP' never propagates into the organizational context where it becomes sticky. Each generation of developers discovers the danger independently. The fix isn't reacting to CVEs; it's treating every file upload endpoint as a critical attack surface from the moment a requirement enters the backlog, not after a vulnerability drops.

Reviewed through automated stages and approved by a human before publication.