CWE-1426
Improper Validation of Generative AI Output
The product invokes a generative AI/ML component whose behaviors and outputs cannot be directly controlled, but the product does not validate or insufficiently validates the outputs to ensure that they align with the intended security, content, or privacy policy.
CVE-2025-31363 (GCVE-0-2025-31363)
Vulnerability from cvelistv5
Published
2025-04-16 09:14
Modified
2025-04-16 14:33
Severity ?
VLAI Severity ?
EPSS score ?
CWE
- CWE-1426 - Improper Validation of Generative AI Output
Summary
Mattermost versions 10.4.x <= 10.4.2, 10.5.x <= 10.5.0, 9.11.x <= 9.11.9 fail to restrict domains the LLM can request to contact upstream which allows an authenticated user to exfiltrate data from an arbitrary server accessible to the victim via performing a prompt injection in the AI plugin's Jira tool.
References
| ► | URL | Tags |
|---|---|---|
Impacted products
| Vendor | Product | Version | ||
|---|---|---|---|---|
| Mattermost | Mattermost |
Version: 10.4.0 ≤ 10.4.2 Version: 10.5.0 Version: 9.11.0 ≤ 9.11.9 |
{
"containers": {
"adp": [
{
"metrics": [
{
"other": {
"content": {
"id": "CVE-2025-31363",
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "no"
},
{
"Technical Impact": "partial"
}
],
"role": "CISA Coordinator",
"timestamp": "2025-04-16T14:19:20.968987Z",
"version": "2.0.3"
},
"type": "ssvc"
}
}
],
"providerMetadata": {
"dateUpdated": "2025-04-16T14:33:01.674Z",
"orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"shortName": "CISA-ADP"
},
"title": "CISA ADP Vulnrichment"
}
],
"cna": {
"affected": [
{
"defaultStatus": "unaffected",
"product": "Mattermost",
"vendor": "Mattermost",
"versions": [
{
"lessThanOrEqual": "10.4.2",
"status": "affected",
"version": "10.4.0",
"versionType": "semver"
},
{
"status": "affected",
"version": "10.5.0"
},
{
"lessThanOrEqual": "9.11.9",
"status": "affected",
"version": "9.11.0",
"versionType": "semver"
},
{
"status": "unaffected",
"version": "10.6.0"
},
{
"status": "unaffected",
"version": "10.4.3"
},
{
"status": "unaffected",
"version": "10.5.1"
},
{
"status": "unaffected",
"version": "9.11.10"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "finder",
"value": "Juho Fors\u00e9n"
}
],
"descriptions": [
{
"lang": "en",
"supportingMedia": [
{
"base64": false,
"type": "text/html",
"value": "\u003cp\u003eMattermost versions 10.4.x \u0026lt;= 10.4.2, 10.5.x \u0026lt;= 10.5.0, 9.11.x \u0026lt;= 9.11.9 fail to restrict domains the LLM can request to contact upstream\u0026nbsp;which allows an authenticated user to\u0026nbsp;\u003cspan style=\"background-color: rgb(255, 255, 255);\"\u003eexfiltrate data from an arbitrary server accessible to the victim via\u003c/span\u003e performing a prompt injection\u0026nbsp;in the AI plugin\u0027s Jira tool.\u003c/p\u003e"
}
],
"value": "Mattermost versions 10.4.x \u003c= 10.4.2, 10.5.x \u003c= 10.5.0, 9.11.x \u003c= 9.11.9 fail to restrict domains the LLM can request to contact upstream\u00a0which allows an authenticated user to\u00a0exfiltrate data from an arbitrary server accessible to the victim via performing a prompt injection\u00a0in the AI plugin\u0027s Jira tool."
}
],
"metrics": [
{
"cvssV3_1": {
"attackComplexity": "HIGH",
"attackVector": "NETWORK",
"availabilityImpact": "NONE",
"baseScore": 3,
"baseSeverity": "LOW",
"confidentialityImpact": "LOW",
"integrityImpact": "NONE",
"privilegesRequired": "LOW",
"scope": "CHANGED",
"userInteraction": "REQUIRED",
"vectorString": "CVSS:3.1/AV:N/AC:H/PR:L/UI:R/S:C/C:L/I:N/A:N",
"version": "3.1"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-1426",
"description": "CWE-1426: Improper Validation of Generative AI Output",
"lang": "en",
"type": "CWE"
}
]
}
],
"providerMetadata": {
"dateUpdated": "2025-04-16T09:14:15.992Z",
"orgId": "9302f53e-dde5-4bf3-b2f2-a83f91ac0eee",
"shortName": "Mattermost"
},
"references": [
{
"url": "https://mattermost.com/security-updates"
}
],
"solutions": [
{
"lang": "en",
"supportingMedia": [
{
"base64": false,
"type": "text/html",
"value": "\u003cp\u003eUpdate Mattermost to versions 10.6.0, 10.4.3, 10.5.1, 9.11.10 or higher.\u003c/p\u003e"
}
],
"value": "Update Mattermost to versions 10.6.0, 10.4.3, 10.5.1, 9.11.10 or higher."
}
],
"source": {
"advisory": "MMSA-2024-00401",
"defect": [
"https://mattermost.atlassian.net/browse/MM-61473"
],
"discovery": "INTERNAL"
},
"title": "Data exfiltration via AI plugin Jira tool",
"x_generator": {
"engine": "Vulnogram 0.2.0"
}
}
},
"cveMetadata": {
"assignerOrgId": "9302f53e-dde5-4bf3-b2f2-a83f91ac0eee",
"assignerShortName": "Mattermost",
"cveId": "CVE-2025-31363",
"datePublished": "2025-04-16T09:14:15.992Z",
"dateReserved": "2025-04-08T07:50:19.617Z",
"dateUpdated": "2025-04-16T14:33:01.674Z",
"state": "PUBLISHED"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.1"
}
Mitigation
Phase: Architecture and Design
Description:
- Since the output from a generative AI component (such as an LLM) cannot be trusted, ensure that it operates in an untrusted or non-privileged space.
Mitigation
Phase: Operation
Description:
- Use "semantic comparators," which are mechanisms that provide semantic comparison to identify objects that might appear different but are semantically similar.
Mitigation
Phase: Operation
Description:
- Use components that operate externally to the system to monitor the output and act as a moderator. These components are called different terms, such as supervisors or guardrails.
Mitigation
Phase: Build and Compilation
Description:
- During model training, use an appropriate variety of good and bad examples to guide preferred outputs.
No CAPEC attack patterns related to this CWE.