FKIE_CVE-2025-12058

Vulnerability from fkie_nvd - Published: 2025-10-29 09:15 - Updated: 2026-10-19 13:10
Severity
5.9 (Medium) - CVSS:4.0/AV:A/AC:H/AT:P/PR:L/UI:P/VC:H/VI:L/VA:L/SC:H/SI:L/SA:L
Summary
The Keras.Model.load_model method, including when executed with the intended security mitigation safe_mode=True, is vulnerable to arbitrary local file loading and Server-Side Request Forgery (SSRF). This vulnerability stems from the way the StringLookup layer is handled during model loading from a specially crafted .keras archive. The constructor for the StringLookup layer accepts a vocabulary argument that can specify a local file path or a remote file path. * Arbitrary Local File Read: An attacker can create a malicious .keras file that embeds a local path in the StringLookup layer's configuration. When the model is loaded, Keras will attempt to read the content of the specified local file and incorporate it into the model state (e.g., retrievable via get_vocabulary()), allowing an attacker to read arbitrary local files on the hosting system. * Server-Side Request Forgery (SSRF): Keras utilizes tf.io.gfile for file operations. Since tf.io.gfile supports remote filesystem handlers (such as GCS and HDFS) and HTTP/HTTPS protocols, the same mechanism can be leveraged to fetch content from arbitrary network endpoints on the server's behalf, resulting in an SSRF condition. The security issue is that the feature allowing external path loading was not properly restricted by the safe_mode=True flag, which was intended to prevent such unintended data access.
References
cve-coordination@google.comhttps://github.com/keras-team/keras/pull/21751
cve-coordination@google.comhttps://github.com/keras-team/keras/security/advisories/GHSA-qg93-c7p6-gg7f
Impacted products
Vendor Product Version
To clipboard 

{
  "affected": [
    {
      "affectedData": [
        {
          "defaultStatus": "unaffected",
          "product": "Keras",
          "vendor": "Keras",
          "versions": [
            {
              "lessThan": "3.12.0",
              "status": "affected",
              "version": "0",
              "versionType": "date"
            }
          ]
        }
      ],
      "source": "cve-coordination@google.com"
    }
  ],
  "cveTags": [],
  "descriptions": [
    {
      "lang": "en",
      "value": "The Keras.Model.load_model method, including when executed with the intended security mitigation safe_mode=True, is vulnerable to arbitrary local file loading and Server-Side Request Forgery (SSRF).\n\n\nThis vulnerability stems from the way the StringLookup layer is handled during model loading from a specially crafted .keras archive. The constructor for the StringLookup layer accepts a vocabulary argument that can specify a local file path or a remote file path.\n\n  *  Arbitrary Local File Read: An attacker can create a malicious .keras file that embeds a local path in the StringLookup layer\u0027s configuration. When the model is loaded, Keras will attempt to read the content of the specified local file and incorporate it into the model state (e.g., retrievable via get_vocabulary()), allowing an attacker to read arbitrary local files on the hosting system.\n\n\n  *  Server-Side Request Forgery (SSRF): Keras utilizes tf.io.gfile for file operations. Since tf.io.gfile supports remote filesystem handlers (such as GCS and HDFS) and HTTP/HTTPS protocols, the same mechanism can be leveraged to fetch content from arbitrary network endpoints on the server\u0027s behalf, resulting in an SSRF condition.\n\n\nThe security issue is that the feature allowing external path loading was not properly restricted by the safe_mode=True flag, which was intended to prevent such unintended data access."
    },
    {
      "lang": "es",
      "value": "El m\u00e9todo Keras.Model.load_model, incluso cuando se ejecuta con la mitigaci\u00f3n de seguridad prevista safe_mode=True, es vulnerable a la carga arbitraria de archivos locales y a la falsificaci\u00f3n de petici\u00f3n del lado del servidor (SSRF).\n\nEsta vulnerabilidad se deriva de la forma en que se maneja la capa StringLookup durante la carga del modelo desde un archivo .keras especialmente dise\u00f1ado. El constructor de la capa StringLookup acepta un argumento vocabulary que puede especificar una ruta de archivo local o una ruta de archivo remota.\n\n*   Lectura Arbitraria de Archivos Locales: Un atacante puede crear un archivo .keras malicioso que incrusta una ruta local en la configuraci\u00f3n de la capa StringLookup. Cuando se carga el modelo, Keras intentar\u00e1 leer el contenido del archivo local especificado e incorporarlo al estado del modelo (por ejemplo, recuperable a trav\u00e9s de get_vocabulary()), permitiendo a un atacante leer archivos locales arbitrarios en el sistema anfitri\u00f3n.\n\n*   Falsificaci\u00f3n de Petici\u00f3n del Lado del Servidor (SSRF): Keras utiliza tf.io.gfile para operaciones de archivos. Dado que tf.io.gfile soporta manejadores de sistemas de archivos remotos (como GCS y HDFS) y protocolos HTTP/HTTPS, el mismo mecanismo puede ser aprovechado para obtener contenido de puntos finales de red arbitrarios en nombre del servidor, lo que resulta en una condici\u00f3n de SSRF.\n\nEl problema de seguridad es que la caracter\u00edstica que permite la carga de rutas externas no fue restringida adecuadamente por el indicador safe_mode=True, el cual estaba destinado a prevenir dicho acceso no intencionado a los datos."
    }
  ],
  "id": "CVE-2025-12058",
  "lastModified": "2026-10-19T13:10:00.110",
  "metrics": {
    "cvssMetricV40": [
      {
        "cvssData": {
          "Automatable": "NOT_DEFINED",
          "Recovery": "NOT_DEFINED",
          "Safety": "NOT_DEFINED",
          "attackComplexity": "HIGH",
          "attackRequirements": "PRESENT",
          "attackVector": "ADJACENT",
          "availabilityRequirement": "NOT_DEFINED",
          "baseScore": 5.9,
          "baseSeverity": "MEDIUM",
          "confidentialityRequirement": "NOT_DEFINED",
          "exploitMaturity": "NOT_DEFINED",
          "integrityRequirement": "NOT_DEFINED",
          "modifiedAttackComplexity": "NOT_DEFINED",
          "modifiedAttackRequirements": "NOT_DEFINED",
          "modifiedAttackVector": "NOT_DEFINED",
          "modifiedPrivilegesRequired": "NOT_DEFINED",
          "modifiedSubAvailabilityImpact": "NOT_DEFINED",
          "modifiedSubConfidentialityImpact": "NOT_DEFINED",
          "modifiedSubIntegrityImpact": "NOT_DEFINED",
          "modifiedUserInteraction": "NOT_DEFINED",
          "modifiedVulnAvailabilityImpact": "NOT_DEFINED",
          "modifiedVulnConfidentialityImpact": "NOT_DEFINED",
          "modifiedVulnIntegrityImpact": "NOT_DEFINED",
          "privilegesRequired": "LOW",
          "providerUrgency": "NOT_DEFINED",
          "subAvailabilityImpact": "LOW",
          "subConfidentialityImpact": "HIGH",
          "subIntegrityImpact": "LOW",
          "userInteraction": "PASSIVE",
          "valueDensity": "NOT_DEFINED",
          "vectorString": "CVSS:4.0/AV:A/AC:H/AT:P/PR:L/UI:P/VC:H/VI:L/VA:L/SC:H/SI:L/SA:L/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",
          "version": "4.0",
          "vulnAvailabilityImpact": "LOW",
          "vulnConfidentialityImpact": "HIGH",
          "vulnIntegrityImpact": "LOW",
          "vulnerabilityResponseEffort": "NOT_DEFINED"
        },
        "source": "cve-coordination@google.com",
        "type": "Secondary"
      }
    ],
    "ssvcV203": [
      {
        "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
        "ssvcData": {
          "id": "CVE-2025-12058",
          "options": [
            {
              "exploitation": "none"
            },
            {
              "automatable": "no"
            },
            {
              "technicalImpact": "partial"
            }
          ],
          "role": "CISA Coordinator",
          "timestamp": "2025-10-29T14:07:04.803189Z",
          "version": "2.0.3"
        }
      }
    ]
  },
  "published": "2025-10-29T09:15:35.500",
  "references": [
    {
      "source": "cve-coordination@google.com",
      "url": "https://github.com/keras-team/keras/pull/21751"
    },
    {
      "source": "cve-coordination@google.com",
      "url": "https://github.com/keras-team/keras/security/advisories/GHSA-qg93-c7p6-gg7f"
    }
  ],
  "sourceIdentifier": "cve-coordination@google.com",
  "vulnStatus": "Deferred",
  "weaknesses": [
    {
      "description": [
        {
          "lang": "en",
          "value": "CWE-502"
        }
      ],
      "source": "cve-coordination@google.com",
      "type": "Secondary"
    }
  ]
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.




Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.

Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline

Sightings and recorded dates, grouped by day · UTC

JSON

Loading timeline…




    Related by attack behaviour

    Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.