Automated Threat Modeling

Brief Summary: SmarterMail CVE-2026-126969, an Authentication Bypass via Low-Entropy Application Passwords

A brief summary of CVE-2026-126969, a CVSS 9.1 authentication bypass in SmarterTools SmarterMail caused by insufficient entropy in Application Password generation. It covers the root cause, the Build 9735 fix and its forced App Password regeneration, affected versions, and vendor history.

CVE Analysis

8 min read

Gaurav Sarraf
Gaurav Sarraf

2026-11-17

Brief Summary: SmarterMail CVE-2026-126969, an Authentication Bypass via Low-Entropy Application Passwords

Introduction

Application Passwords let desktop and mobile mail clients keep working on accounts protected by two-factor authentication, so their security depends entirely on how hard they are to guess. In SmarterMail, a weakness in how these credentials were generated let remote, unauthenticated attackers bypass authentication. The issue carries a CVSS score of 9.1.

SmarterMail is an email, groupware, and collaboration server from SmarterTools, a Phoenix, Arizona company founded in 2003 that originally built software for web hosting companies. It bundles email, shared calendars, contacts, tasks, notes, group chat, and online meetings, runs on Linux, Windows, and Docker, and is marketed to enterprises, hosting providers, ISPs, and smaller businesses as a Microsoft Exchange alternative. SmarterTools reports more than 15 million users in more than 120 countries across its products. These are company-wide vendor claims, not independent counts of SmarterMail deployments.

Technical information

Root cause: weak credential generation

The Zero Day Initiative published this issue as ZDI-26-997. It was tracked earlier as ZDI-CAN-33568 and credited to Shaber Tseng. The advisory places the flaw in the generation of Application Passwords, which used insufficient entropy. The GitHub advisory GHSA-5frr-33p9-gf27 uses the same classification, CWE-331: Insufficient Entropy.

MITRE describes CWE-331 as a scheme whose output does not have enough entropy, so some patterns or clusters of values are more likely than others. When those values serve as authentication material, an attacker who can guess them gets unauthorized access. The flaw is in how the product generates credentials. It has nothing to do with users choosing weak passwords, so password policy changes do not reach it.

Why App Passwords amplify the problem

In SmarterMail, App Passwords are per-device credentials. They let desktop and mobile clients sign in to 2FA-enabled accounts without a second factor. Because they skip 2FA by design, their only protection is unpredictability. If the generator produces values from a small or skewed space, an attacker can predict or enumerate likely candidates. A correct guess signs the attacker in as the user, without the real account password and without the second factor.

This also explains why turning on 2FA does not help here. The vulnerable credential type exists specifically to work alongside 2FA, and the Build 9735 release note is written for clients accessing 2FA-enabled accounts.

What is not disclosed

SmarterMail is closed source, and no public commit or diff exists. The ZDI advisory and the NVD record do not disclose any of the following:

  • the random number generator used
  • how the generator was seeded
  • the effective entropy
  • the credential length or character set
  • the endpoint or protocol involved
  • the number of attempts required
  • a reproducible exploit sequence

In .NET, CWE-331 is usually fixed by replacing a non-cryptographic or time-seeded random generator with a cryptographically secure one and lengthening the output. SmarterTools has not confirmed that this is what happened in SmarterMail. Treat it as general guidance only.

Severity vector

NVD shows the following CVSS v3.0 vector, supplied by ZDI as the reporting source. In the record we reviewed, NVD had not yet published its own assessment.

CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N
  • Attack Vector: Network, so the attacker can work remotely.
  • Attack Complexity: Low. This is a published rating, not a measured time to exploit.
  • Privileges Required: None. The attacker needs no existing account.
  • User Interaction: None.
  • Scope: Unchanged.
  • Confidentiality and Integrity: High.
  • Availability: None.

The High confidentiality and integrity ratings fit an attacker reading and changing a mailbox owner's data after a successful sign in. The record does not establish remote code execution, operating system compromise, administrator impersonation, or cross-system compromise as consequences of this CVE.

Attack flow

The advisory and patch notes support the following general sequence, though not its exact mechanics.

  1. The attacker reaches a SmarterMail server over the network. No credentials or victim interaction are needed.
  2. The target is an account with App Passwords issued by the generator used before Build 9735.
  3. Because that generator had insufficient entropy, its outputs fall into a predictable or enumerable set. The attacker predicts or enumerates candidate App Password values.
  4. The attacker submits a candidate where SmarterMail accepts App Passwords for client sign in. A correct value authenticates the attacker as the user, without the primary password and without 2FA.
  5. The attacker then has authenticated access to that account. The vector rates the resulting confidentiality and integrity impact High and the availability impact None.

Exploitation status

The reviewed trackers (TheHackerWire, OffSeq Threat Radar) report no confirmed public exploit, no known exploitation in the wild, and no CISA KEV listing. OffSeq found no exploit indexed in Metasploit or Exploit DB. TheHackerWire lists the EPSS score as still being calculated. No threat actor has been linked to this CVE. These findings reflect public evidence only and do not rule out private exploitation.

Patch information

Fixed build. SmarterTools fixed CVE-2026-126969 in SmarterMail Build 9735, released August 27, 2026. ZDI's advisory states that the issue is "Fixed in SmarterMail Build 9735". ZDI's timeline records that the bug was reported to the vendor on August 20, 2026, so the fix shipped about a week later. The advisory became public on November 13, 2026, and NVD published on November 18, 2026. Patched builds were therefore available for roughly 11 weeks before details were released.

How the fix works. The first entry in the Build 9735 release notes is marked IMPORTANT. It states that App Passwords "were regenerated to a new security scheme." Anyone using an App Password created before Build 9735 must enter a new one before a desktop or mobile client can use a 2FA-enabled account again. The fix has two parts:

  1. A new generation scheme. From Build 9735 on, App Passwords come from a different generator. SmarterTools has not published its random source, length, or character set.
  2. Forced regeneration of every existing App Password. This is the part that matters more. A new generator only protects credentials created after the upgrade, and any App Password issued by the old generator would remain guessable and accepted. Discarding all existing App Passwords during the upgrade ensures no low-entropy credential stays valid on the server. The cost is that every user has to set up their mail clients again.

Earlier changes that did not fix this. App Password handling changed twice before, but neither change touched generation:

  • Build 9406 (October 2025): App Passwords were regenerated automatically whenever a user changed their main password, which is rotation.
  • Build 9581 (March 2026): App Passwords were stored with AES, which is protection at rest.

A credential that was predictable when it was created stays predictable after it is encrypted. Build 9735 is the first release to change how the values are created.

Unrelated authentication fixes in the same build. The Build 9735 notes also list several other fixes. None of them is tied to this CVE:

  • login failures that could allow user enumeration
  • authentication endpoints returning a refreshToken when no clientId was supplied
  • password reset links that were not reliably expired after use

Upgrade expectations. Install Build 9735 or any later build. The current release is Build 9817, released November 17, 2026. On upgrade, the server invalidates old App Passwords automatically. Expect desktop and mobile clients on 2FA-enabled accounts to fail sign in until users generate and enter new App Passwords. Those failures mean the fix is working, and they do not indicate a failed upgrade. ZDI documents no configuration workaround. Firewall rules, rate limits, and password policy changes have not been verified as substitutes for the update.

Affected systems and versions

  • Product: SmarterTools SmarterMail
  • Listed as affected by NVD: SmarterMail 100.0.9693.29286
  • Fixed: Build 9735 and later, with Build 9817 as the current release
  • Exposed configuration: accounts with App Passwords generated before Build 9735, which are the credentials desktop and mobile clients use to access 2FA-enabled accounts

The reviewed sources do not give a complete affected version range. Treat any build earlier than 9735 as requiring the update.

Vendor security history

SmarterMail has had several serious issues in the past year. Each has its own mechanism, remediation, and evidence base:

  • CVE-2025-52691: According to the Cyber Security Agency of Singapore, this arbitrary file upload flaw affects Build 9406 and earlier and could enable remote code execution. The agency directs users to Build 9413.
  • CVE-2026-23760: NVD describes an anonymous password reset API that did not verify the existing password or reset token for system administrator accounts. It led to administrative compromise and possible operating system command execution, and its fix boundary is Build 9511. NVD records that CISA added it to KEV on January 26, 2026.
  • SmarterTools network breach: SmarterTools reported a breach of its own network on January 29, 2026. Entry came through an employee-created VM that had not been updated, and about 12 Windows servers appeared compromised. The vendor says business applications and account data were not affected and attributes the activity to the Warlock Group. These are the company's own findings. The incident reflects a gap in deploying available fixes and is not evidence of exploitation of CVE-2026-126969.

The exploitation evidence and broader impact documented for CVE-2026-23760 do not carry over to this CVE. CVE-2026-126969 is rated critical, but public exploitation is currently unconfirmed.

References

Detect & fix
what others miss

Works with
  • GitHub
  • GitLab
  • Bitbucket
  • Azure DevOps Services
  • Gerrit
  • CVS
  • Jira
  • Linear
  • Slack
  • Security Compass
Explore integrations
Security magnifying glass visualization