s2
--:--:--UTC

Searching APEX

Starting…

  1. Searching Threats, IOCs & Threat Intelligence locally
  2. Querying external providers
  3. Asking AI Forensic Validator
  4. Creating new entry from validated hit

0s elapsed

CVE-2026-31239

📛 CVE Title

CVE-2026-31239

Description

The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization (CWE-502) when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.from_pretrained() method uses torch.load() to load the pytorch_model.bin weight file without enabling the security-restrictive weights_only=True parameter. This allows the deserialization of arbitrary Python objects via the pickle module. An attacker can exploit this by publishing a malicious model repository on HuggingFace Hub. When a victim loads a model from this repository, arbitrary code is executed on the victim's system in the context of the mamba process.

Overview

State
PUBLISHED
Assigner (CNA)
mitre
CVSS severity
critical
CVSS score
CVSS 9.8 / 10 9.8 9.8 / 10
CVSS vector
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Effective score
9.8 / 10 CRITICAL source: CNA overview
CWE(s)
Reserved
2026-03-09
Published
2026-05-12 00:00 UTC
Last updated
2026-05-14 19:54 UTC
Source
https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/31xxx/CVE-2026-31239.json
Linked Threat
CVE-2026-31239 — CVE-2026-31239

NVD triage scoring NVD CVE 2.0

Layer NVD adds on top of the CNA's CVE record — published / last-modified timestamps, exploitability / impact subscores, and the FIRST.org EPSS probability that this CVE will be exploited in the wild in the next 30 days.

NVD published
2026-05-12 18:16:52 UTC
NVD last modified
2026-05-14 20:17:03 UTC
NVD CVSS v3.1
CVSS 9.8 / 10 9.8 9.8 / 10 CRITICAL source: 134c704f-9b21-4f2e-91b3-4a467353bcc0
NVD CVSS vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Exploitability subscore
3.9 / 10
Impact subscore
5.9 / 10
EPSS score
0.0005 (probability of exploitation in next 30 days)
EPSS percentile
16.93% vs all CVEs — higher = more likely to be exploited, as of 2026-05-24

NVD-assigned CWE(s): CWE-502 (differs from the CNA list above)

NVD / KEV / EPSS data refreshed 2026-05-25 03:34 UTC. Re-run the 🛰 Backfill from NVD button above to refresh.

European Union Vulnerability Database ENISA EUVD

ENISA's official EU repository for curated vulnerability intelligence. Carries a separate identifier (EUVD-YYYY-NNNN) and frequently exposes an earlier-published description + CVSS than NVD does.

EUVD ID
EUVD-2026-29562
Assigner
mitre
Published
May 12, 2026, 12:00:00 AM
Updated
May 14, 2026, 7:54:17 PM
EUVD base score (CVSS 3.1)
9.8 / 10
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
EUVD-reported EPSS
0.0500
Vendors
n/a
Products
n/a (n/a)

ENISA description: The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization (CWE-502) when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.from_pretrained() method uses torch.load() to load the pytorch_model.bin weight file without enabling the security-restrictive weights_only=True parameter. This allows the deserialization of arbitrary Python objects via the pickle module. An attacker can exploit this by publishing a malicious model repository on HuggingFace Hub. When a victim loads a model from this repository, arbitrary code is executed on the victim's system in the context of the mamba process.

EUVD references (2)

Affected products (1)

VendorProductVersionsPlatforms
n/a n/a n/a (affected)

Vendor references (2)

References embedded in the original CVE record by the assigning CNA.

Web references (5)

DuckDuckGo results ranked by threat-intel / vendor advisory domains. Generated by the 🔎 Find references (web) button above — same flow as the Remediations search.

NVD-tagged references (2)

Reference list NVD curates from the CNA record, vendor advisories, and third-party reports. The tag chips below are NVD's analyst-assigned categories.

Remediations (14)

Remediations are stored against the linked Threat row; the list below is deduplicated across both pages.

  • web:blog.qualys.com

    With Qualys Policy Audit's out-of-the-box mitigation or Compensatory Controls, which reduce the risk of a vulnerability being exploited because the remediation ( fix / patch ) cannot be done immediately, these security controls are not recommended by any industry standards, such as CIS and DISA-STIG.

    2026-05-23 21:46 UTC
  • web:cyberpress.org

    March 2026 Patch Tuesday Overview According to Microsoft's security advisory, the March release fixes a total of 79 vulnerabilities across the Microsoft ecosystem.

    2026-05-23 21:46 UTC
  • web:cybersecuritynews.com

    Microsoft has released its April 2026 Patch Tuesday security update, addressing 168 vulnerabilities across its product portfolio including one actively exploited zero-day and one publicly disclosed flaw that organizations must prioritize immediately.

    2026-05-23 21:46 UTC
  • web:krebsonsecurity.com

    Microsoft expects that exploitation is more likely. May's Patch Tuesday is a welcome respite from April, which saw Microsoft fix a near-record 167 security flaws.

    2026-05-23 21:46 UTC
  • web:portal.msrc.microsoft.com

    The Security Update Guide provides information on the latest Microsoft security updates, helping users understand and address potential vulnerabilities effectively.

    2026-05-23 21:46 UTC
  • web:www.crowdstrike.com

    Microsoft's April 2026 Patch Tuesday addresses 164 CVEs , featuring 8 Critical vulnerabilities, one exploited zero-day, and one disclosed zero-day.

    2026-05-23 21:46 UTC
  • web:www.oracle.com

    This Critical Patch Update contains 481 new security patches across the product families listed below. Please note that an MOS note summarizing the content of this Critical Patch Update and other Oracle Software Security Assurance activities is located at April 2026 Critical Patch Update: Executive Summary and Analysis.

    2026-05-23 21:46 UTC
  • web:www.sentinelone.com

    CVE - 2026 -41239 is an XSS vulnerability in DOMPurify sanitizer. Learn about its impact, affected versions, and mitigation methods.

    2026-05-23 21:46 UTC
  • web:www.windowslatest.com

    Windows 11 April 2026 update adds Narrator Copilot support, faster Settings, File Explorer fixes, and key security improvements.

    2026-05-23 21:46 UTC
  • web:zecurit.com

    Get the complete breakdown of Microsoft's May 2026 Patch Tuesday. We analyze the latest security updates and all critical CVEs .

    2026-05-23 21:46 UTC
  • web:krebsonsecurity.com

    For a clickable, per- patch breakdown, check out the SANS Internet Storm Center Patch Tuesday roundup. Running into problems applying any of these updates?

    2026-05-26 02:51 UTC
  • web:msrc.microsoft.com

    The Microsoft Security Response Center (MSRC) investigates all reports of security vulnerabilities affecting Microsoft products and services, and provides the information here as part of the ongoing effort to help you manage security risks and help keep your systems protected.

    2026-05-26 02:51 UTC
  • web:www.bleepingcomputer.com

    Today is Microsoft's April 2026 Patch Tuesday with security updates for 167 flaws, including 2 zero-day vulnerabilities.

    2026-05-26 02:51 UTC
  • web:www.pcworld.com

    This month's Patch Tuesday includes an actively exploited Office zero-day vulnerability and several critical RCE bugs in Windows and Remote Desktop.

    2026-05-26 02:51 UTC

AI Forensic Analysis

Only Available for Registered Users. Sign in to view.

Raw JSON

The full cvelistV5 record. Download as CVE-2026-31239.json.

{
  "containers": {
    "adp": [
      {
        "metrics": [
          {
            "cvssV3_1": {
              "attackComplexity": "LOW",
              "attackVector": "NETWORK",
              "availabilityImpact": "HIGH",
              "baseScore": 9.8,
              "baseSeverity": "CRITICAL",
              "confidentialityImpact": "HIGH",
              "integrityImpact": "HIGH",
              "privilegesRequired": "NONE",
              "scope": "UNCHANGED",
              "userInteraction": "NONE",
              "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
              "version": "3.1"
            }
          },
          {
            "other": {
              "content": {
                "id": "CVE-2026-31239",
                "options": [
                  {
                    "Exploitation": "none"
                  },
                  {
                    "Automatable": "yes"
                  },
                  {
                    "Technical Impact": "total"
                  }
                ],
                "role": "CISA Coordinator",
                "timestamp": "2026-05-14T16:52:22.756950Z",
                "version": "2.0.3"
              },
              "type": "ssvc"
            }
          }
        ],
        "problemTypes": [
          {
            "descriptions": [
              {
                "cweId": "CWE-502",
                "description": "CWE-502 Deserialization of Untrusted Data",
                "lang": "en",
                "type": "CWE"
              }
            ]
          }
        ],
        "providerMetadata": {
          "dateUpdated": "2026-05-14T19:54:17.202Z",
          "orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
          "shortName": "CISA-ADP"
        },
        "title": "CISA ADP Vulnrichment"
      }
    ],
    "cna": {
      "affected": [
        {
          "product": "n/a",
          "vendor": "n/a",
          "versions": [
            {
              "status": "affected",
              "version": "n/a"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization (CWE-502) when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.from_pretrained() method uses torch.load() to load the pytorch_model.bin weight file without enabling the security-restrictive weights_only=True parameter. This allows the deserialization of arbitrary Python objects via the pickle module. An attacker can exploit this by publishing a malicious model repository on HuggingFace Hub. When a victim loads a model from this repository, arbitrary code is executed on the victim's system in the context of the mamba process."
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "description": "n/a",
              "lang": "en",
              "type": "text"
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-05-12T17:16:58.846Z",
        "orgId": "8254265b-2729-46b6-b9e3-3dfca2d5bfca",
        "shortName": "mitre"
      },
      "references": [
        {
          "url": "https://github.com/state-spaces/mamba"
        },
        {
          "url": "https://www.notion.so/CVE-2026-31239-35d1e1393188810d9baedfbd8363f396"
        }
      ]
    }
  },
  "cveMetadata": {
    "assignerOrgId": "8254265b-2729-46b6-b9e3-3dfca2d5bfca",
    "assignerShortName": "mitre",
    "cveId": "CVE-2026-31239",
    "datePublished": "2026-05-12T00:00:00.000Z",
    "dateReserved": "2026-03-09T00:00:00.000Z",
    "dateUpdated": "2026-05-14T19:54:17.202Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2"
}