CVE-2026-73576: 
Zimbra Collaboration Server vulnerability analysis and mitigation

CVE-2026-73576 is a weak cryptographic key generation vulnerability in the OnlyOffice integration of Zimbra Collaboration (ZCS). The zimbraDocumentEditingJwtSecret is generated using an insecure (predictable) random number generator with insufficient entropy, allowing an attacker who obtains a signed JWT to recover the signing secret via offline brute-force and subsequently forge JWTs. All ZCS versions before 10.1.17 are affected. It carries a CVSS v3.1 base score of 6.3 (Medium), published on August 13, 2026 (Github Advisory, Zimbra Security Advisories).

The root cause is classified as CWE-1241 (Use of Predictable Algorithm in Random Number Generator). The zimbraDocumentEditingJwtSecret, used to sign JWTs for the OnlyOffice document editing integration, is derived from a non-cryptographically-secure pseudo-random number generator (PRNG), resulting in a low-entropy secret that is susceptible to offline brute-force recovery. An attacker with low-privilege network access who can obtain a valid JWT (e.g., by authenticating as a regular user and capturing the token) can then attempt to crack the signing secret offline and craft forged JWTs. The attack is network-based, requires low privileges, and no user interaction, but has high attack complexity due to the offline brute-force requirement (Github Advisory, Zimbra Security Advisories).

Successful exploitation allows an authenticated attacker to forge JWT tokens, enabling impersonation of other users — including potentially administrators — within the OnlyOffice document editing functionality of Zimbra Collaboration. The primary impact is a high integrity loss (unauthorized modification of authentication state and document access), with no direct confidentiality or availability impact per the CVSS scoring. The changed scope indicates that the impact extends beyond the vulnerable component itself, potentially affecting other users' document sessions and data (Github Advisory).

There is no public proof-of-concept exploit and no evidence of in-the-wild exploitation at this time (Github Advisory). The EPSS score is approximately 0.188% (8th percentile), indicating a low near-term exploitation probability. The vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog. NVD's SSVC assessment classifies exploitation as "none" and the attack as non-automatable, reflecting the high complexity of the offline brute-force step required.

1. Authenticate as a low-privilege user: Log in to the target Zimbra Collaboration instance with any valid user account to gain access to the OnlyOffice document editing feature.
2. Capture a signed JWT: Open or interact with a document in the OnlyOffice integration to trigger JWT issuance. Intercept the JWT (e.g., via browser developer tools or a proxy such as Burp Suite) from the HTTP request/response to the OnlyOffice service.
3. Offline brute-force the JWT secret: Use a tool such as hashcat or jwt-cracker to perform an offline dictionary or brute-force attack against the captured JWT's HMAC signature, exploiting the low entropy of the zimbraDocumentEditingJwtSecret.
4. Forge a JWT: Once the secret is recovered, craft a new JWT with a modified sub or user claim (e.g., targeting an administrator account) using a library such as python-jwt or jsonwebtoken.
5. Access document editing as the impersonated user: Submit the forged JWT to the OnlyOffice integration endpoint to gain unauthorized access to documents or sessions belonging to the impersonated user (Github Advisory).

• Network: Unusual or repeated requests to OnlyOffice integration endpoints (/service/extension/onlyoffice/ or similar) with JWT tokens from unexpected source IPs or user agents.
• Logs: Zimbra mailbox logs showing document editing sessions initiated by a user account from an IP address inconsistent with that user's normal activity; multiple JWT validation attempts from the same source.
• Application: Evidence of the same document being accessed by multiple distinct user identities in a short timeframe, particularly if one identity is an administrator.
• Configuration: Presence of a low-entropy or short zimbraDocumentEditingJwtSecret value in Zimbra configuration (detectable via zmprov gcf zimbraDocumentEditingJwtSecret).

Upgrade Zimbra Collaboration to version 10.1.17 or later, which addresses the weak key generation by using a cryptographically secure random number generator for zimbraDocumentEditingJwtSecret (Zimbra Security Advisories, Github Advisory). After upgrading, administrators should rotate the existing zimbraDocumentEditingJwtSecret to ensure any previously generated weak secrets are replaced with a high-entropy value. This can be done via the Zimbra CLI: zmprov mcf zimbraDocumentEditingJwtSecret <new-strong-secret>. If immediate upgrade is not possible, consider disabling the OnlyOffice document editing integration until patching is feasible.

The vulnerability was disclosed by Zimbra on August 13, 2026, with updates to the Zimbra Security Advisories wiki and the 10.1.17 release notes (Zimbra Security Advisories). Coverage has been limited to automated vulnerability tracking platforms such as VulDB, CVEFeed, and INCIBE-CERT, with no notable independent researcher commentary or significant social media discussion observed at this time.