Which versions of GROWI are affected by CVE-2026-25083?

GROWI v7.4.5 and earlier expose every organization's confidential AI-assisted communications to unauthorized access and modification by any authenticated employee through an unprotected API layer.

How to fix or mitigate CVE-2026-25083?

Within 24 hours: Identify all GROWI instances running v7.4.5 or earlier, review API access logs for suspicious activity on AI assistant endpoints, and notify users of the exposure window. Within 7 days: Disable or restrict the AI assistant feature to critical operations only, implement compensating controls (see below), and assess scope of potential unauthorized access. Within 30 days: Monitor vendor advisories and plan immediate upgrade to patched version once available; conduct security audit of access controls across the AI assistant layer.

GROWI CVE-2026-25083

Q: What is the CVSS score of CVE-2026-25083?

CVE-2026-25083 has a CVSS 4.0 base score of 8.7 (High). CVSS vector: CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:L/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X. EPSS exploitation probability: 0.0%.

EUVD-2026-12343 HIGH

Missing Authorization (CWE-862)

2026-03-16 jpcert

Authentication Bypass Growi

8.7

CVSS 4.0 · NVD

Severity by source

NVD PRIMARY

8.7 HIGH

CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:L/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Primary rating from NVD · only source for this CVE.

CVSS VectorNVD

CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:L/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Attack Vector

Network

Attack Complexity

Low

Privileges Required

Low

User Interaction

None

Scope

Lifecycle Timeline

Analysis Updated

May 19, 2026 - 15:43 vuln.today

v1 (cvss_changed)

Re-analysis Queued

May 19, 2026 - 15:37 vuln.today

cvss_changed

CVSS changed

May 19, 2026 - 15:37 NVD

8.3 (HIGH) 8.7 (HIGH)

EUVD ID Assigned

Mar 16, 2026 - 09:00 euvd

EUVD-2026-12343

Analysis Generated

Mar 16, 2026 - 09:00 vuln.today

CVE Published

Mar 16, 2026 - 06:47 nvd

HIGH 8.3

DescriptionCVE.org

GROWI OpenAI thread/message API endpoints do not perform authorization. Affected are v7.4.5 and earlier versions. A logged-in user who knows a shared AI assistant's identifier may view and/or tamper the other user's threads/messages.

AnalysisAI

Cross-user data exposure and tampering in GROWI v7.4.5 and earlier allows any authenticated user who learns a shared AI assistant's identifier to read or modify other users' OpenAI thread and message contents via API endpoints that omit authorization checks. The flaw stems from missing access control (CWE-862) on the AI assistant interaction layer rather than a memory-corruption or injection bug. No public exploit identified at time of analysis, and EPSS rates near-term exploitation probability at 0.04% (12th percentile).

Technical ContextAI

GROWI is an open-source enterprise wiki/collaboration platform developed by WESEEK (growi.co.jp) that integrates an OpenAI-backed AI assistant feature, mirroring the OpenAI Assistants API concept of persistent threads (conversation containers) and messages (individual turns). The vulnerability is a textbook CWE-862 Missing Authorization: the REST endpoints that read and mutate threads/messages authenticate the caller's session but never verify that the caller is the owner of the targeted thread/message resource, so any logged-in account can act on another user's thread by supplying the corresponding shared-assistant identifier. The CPE cpe:2.3:a:growi,_inc.:growi:*:*:*:*:*:*:*:* indicates all version branches up through 7.4.5 are in scope.

RemediationAI

No vendor-released patch version is cited in the provided intelligence, but the vendor has published an advisory at https://growi.co.jp/news/41/ and JPCERT's coordinated disclosure at https://jvn.jp/en/jp/JVN46373837/ - administrators should consult those advisories for the fixed release (expected to be the next version after 7.4.5) and upgrade as soon as it is available. Until a fixed build is deployed, compensating controls include disabling the OpenAI assistant integration in GROWI configuration to remove the vulnerable endpoints from the attack surface (trade-off: loss of AI assistant functionality for all users), restricting access to the GROWI instance to a trusted user pool via SSO/IdP group gating (trade-off: does not stop malicious insiders), and treating shared AI assistant identifiers as sensitive - rotate them and audit logs for unexpected thread/message reads or edits. Generic network filtering will not help because the attack uses authenticated application-layer requests.