CVE-2026-12570 (GCVE-0-2026-12570)

Vulnerability from cvelistv5 – Published: 2026-08-10 06:29 – Updated: 2026-08-10 17:37
VLAI
Title
Denial of Service via HDF5 Shape Bomb in keras.models.load_model() in keras-team/keras
Summary
A vulnerability in keras-team/keras versions <= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function. The H5IOStore.__getitem__ method in keras/src/saving/saving_lib.py does not validate the shape or size of datasets, leading to unbounded memory allocation. A specially crafted .keras file can exploit this flaw to trigger an out-of-memory (OOM) condition, causing the process to be terminated (exit code 137). This issue bypasses the fix for CVE-2026-0897, which only addressed a similar vulnerability in KerasFileEditor. The attack vector includes poisoned models from public repositories or malicious model registries, posing a risk to machine learning pipelines that process untrusted models.
Severity
5.5 (Medium)
CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H
SSVC
Exploitation: poc Automatable: no Technical Impact: partial
CISA Coordinator · CISA-ADP (v2.0.3)
Decision recorded 2026-08-10 17:37 UTC
CWE
  • CWE-770 - Allocation of Resources Without Limits or Throttling
Assigner
@huntr_ai CNA under the mitre root
CNA scorecard C 65/100 over 40 records in the last 180 days details
References
Impacted products
Vendor Product Version CPE status
keras-team keras-team/keras Affected: unspecified , < 3.12.3, 3.15.0 (custom)
guessed Create a notification for this product.
Show details on NVD website
To clipboard 

{
  "containers": {
    "adp": [
      {
        "metrics": [
          {
            "other": {
              "content": {
                "id": "CVE-2026-12570",
                "options": [
                  {
                    "Exploitation": "poc"
                  },
                  {
                    "Automatable": "no"
                  },
                  {
                    "Technical Impact": "partial"
                  }
                ],
                "role": "CISA Coordinator",
                "timestamp": "2026-08-10T17:37:17.637683Z",
                "version": "2.0.3"
              },
              "type": "ssvc"
            }
          }
        ],
        "providerMetadata": {
          "dateUpdated": "2026-08-10T17:37:39.037Z",
          "orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
          "shortName": "CISA-ADP"
        },
        "references": [
          {
            "tags": [
              "exploit"
            ],
            "url": "https://huntr.com/bounties/a064f475-780a-409a-82f7-678512f27ad8"
          }
        ],
        "title": "CISA ADP Vulnrichment"
      }
    ],
    "cna": {
      "affected": [
        {
          "product": "keras-team/keras",
          "vendor": "keras-team",
          "versions": [
            {
              "lessThan": "3.12.3, 3.15.0",
              "status": "affected",
              "version": "unspecified",
              "versionType": "custom"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "A vulnerability in keras-team/keras versions \u003c= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function. The H5IOStore.__getitem__ method in keras/src/saving/saving_lib.py does not validate the shape or size of datasets, leading to unbounded memory allocation. A specially crafted .keras file can exploit this flaw to trigger an out-of-memory (OOM) condition, causing the process to be terminated (exit code 137). This issue bypasses the fix for CVE-2026-0897, which only addressed a similar vulnerability in KerasFileEditor. The attack vector includes poisoned models from public repositories or malicious model registries, posing a risk to machine learning pipelines that process untrusted models."
        }
      ],
      "metrics": [
        {
          "cvssV3_0": {
            "attackComplexity": "LOW",
            "attackVector": "LOCAL",
            "availabilityImpact": "HIGH",
            "baseScore": 5.5,
            "baseSeverity": "MEDIUM",
            "confidentialityImpact": "NONE",
            "integrityImpact": "NONE",
            "privilegesRequired": "NONE",
            "scope": "UNCHANGED",
            "userInteraction": "REQUIRED",
            "vectorString": "CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H",
            "version": "3.0"
          }
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "cweId": "CWE-770",
              "description": "CWE-770 Allocation of Resources Without Limits or Throttling",
              "lang": "en",
              "type": "CWE"
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-08-10T06:29:56.111Z",
        "orgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
        "shortName": "@huntr_ai"
      },
      "references": [
        {
          "url": "https://huntr.com/bounties/a064f475-780a-409a-82f7-678512f27ad8"
        },
        {
          "url": "https://github.com/keras-team/keras/commit/4933ea4a5b3fcc24ceacdc276f5bb5dfbd06756c"
        }
      ],
      "source": {
        "advisory": "a064f475-780a-409a-82f7-678512f27ad8",
        "discovery": "EXTERNAL"
      },
      "title": "Denial of Service via HDF5 Shape Bomb in keras.models.load_model() in keras-team/keras"
    }
  },
  "cveMetadata": {
    "assignerOrgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
    "assignerShortName": "@huntr_ai",
    "cveId": "CVE-2026-12570",
    "datePublished": "2026-08-10T06:29:56.111Z",
    "dateReserved": "2026-06-18T01:55:25.311Z",
    "dateUpdated": "2026-08-10T17:37:39.037Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2",
  "vulnerability-lookup:meta": {
    "epss": {
      "cve": "CVE-2026-12570",
      "date": "2026-11-01",
      "epss": "0.0019",
      "percentile": "0.0873"
    },
    "microsoft_vex": {
      "current_release_date": "2026-08-23T14:57:02.000Z",
      "cve": "CVE-2026-12570",
      "id": "msrc_CVE-2026-12570",
      "initial_release_date": "2026-08-14T19:03:07.000Z",
      "product_status:known_not_affected": "1",
      "source": "Microsoft CSAF VEX",
      "status": "final",
      "title": "Denial of Service via HDF5 Shape Bomb in keras.models.load_model() in keras-team/keras",
      "url": "https://msrc.microsoft.com/csaf/vex/2026/msrc_cve-2026-12570.json",
      "version": "2"
    },
    "nvd": {
      "cve": {
        "affected": [
          {
            "affectedData": [
              {
                "product": "keras-team/keras",
                "vendor": "keras-team",
                "versions": [
                  {
                    "lessThan": "3.12.3, 3.15.0",
                    "status": "affected",
                    "version": "unspecified",
                    "versionType": "custom"
                  }
                ]
              }
            ],
            "source": "security@huntr.dev"
          }
        ],
        "cveTags": [],
        "descriptions": [
          {
            "lang": "en",
            "value": "A vulnerability in keras-team/keras versions \u003c= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function. The H5IOStore.__getitem__ method in keras/src/saving/saving_lib.py does not validate the shape or size of datasets, leading to unbounded memory allocation. A specially crafted .keras file can exploit this flaw to trigger an out-of-memory (OOM) condition, causing the process to be terminated (exit code 137). This issue bypasses the fix for CVE-2026-0897, which only addressed a similar vulnerability in KerasFileEditor. The attack vector includes poisoned models from public repositories or malicious model registries, posing a risk to machine learning pipelines that process untrusted models."
          },
          {
            "lang": "es",
            "value": "Una vulnerabilidad en las versiones menor o igual que 3.15.0 de keras-team/keras permite un ataque de denegaci\u00f3n de servicio (DoS) al cargar archivos de modelo .keras maliciosos a trav\u00e9s de la funci\u00f3n keras.models.load_model(). El m\u00e9todo H5IOStore.__getitem__ en keras/src/saving/saving_lib.py no valida la forma o el tama\u00f1o de los conjuntos de datos, lo que lleva a una asignaci\u00f3n de memoria ilimitada. Un archivo .keras especialmente dise\u00f1ado puede explotar esta falla para desencadenar una condici\u00f3n de falta de memoria (OOM), lo que provoca la terminaci\u00f3n del proceso (c\u00f3digo de salida 137). Este problema elude la correcci\u00f3n para CVE-2026-0897, que solo abord\u00f3 una vulnerabilidad similar en KerasFileEditor. El vector de ataque incluye modelos envenenados de repositorios p\u00fablicos o registros de modelos maliciosos, lo que representa un riesgo para las tuber\u00edas de aprendizaje autom\u00e1tico que procesan modelos no confiables."
          }
        ],
        "id": "CVE-2026-12570",
        "lastModified": "2026-10-23T20:10:00.160",
        "metrics": {
          "cvssMetricV30": [
            {
              "cvssData": {
                "attackComplexity": "LOW",
                "attackVector": "LOCAL",
                "availabilityImpact": "HIGH",
                "baseScore": 5.5,
                "baseSeverity": "MEDIUM",
                "confidentialityImpact": "NONE",
                "integrityImpact": "NONE",
                "privilegesRequired": "NONE",
                "scope": "UNCHANGED",
                "userInteraction": "REQUIRED",
                "vectorString": "CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H",
                "version": "3.0"
              },
              "exploitabilityScore": 1.8,
              "impactScore": 3.6,
              "source": "security@huntr.dev",
              "type": "Secondary"
            }
          ],
          "ssvcV203": [
            {
              "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
              "ssvcData": {
                "id": "CVE-2026-12570",
                "options": [
                  {
                    "exploitation": "poc"
                  },
                  {
                    "automatable": "no"
                  },
                  {
                    "technicalImpact": "partial"
                  }
                ],
                "role": "CISA Coordinator",
                "timestamp": "2026-08-10T17:37:17.637683Z",
                "version": "2.0.3"
              }
            }
          ]
        },
        "published": "2026-08-10T07:16:44.370",
        "references": [
          {
            "source": "security@huntr.dev",
            "url": "https://github.com/keras-team/keras/commit/4933ea4a5b3fcc24ceacdc276f5bb5dfbd06756c"
          },
          {
            "source": "security@huntr.dev",
            "url": "https://huntr.com/bounties/a064f475-780a-409a-82f7-678512f27ad8"
          },
          {
            "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
            "url": "https://huntr.com/bounties/a064f475-780a-409a-82f7-678512f27ad8"
          }
        ],
        "sourceIdentifier": "security@huntr.dev",
        "vulnStatus": "Awaiting Analysis",
        "weaknesses": [
          {
            "description": [
              {
                "lang": "en",
                "value": "CWE-770"
              }
            ],
            "source": "security@huntr.dev",
            "type": "Secondary"
          }
        ]
      }
    },
    "redhat_vex": {
      "aggregate_severity": "Moderate",
      "current_release_date": "2026-08-10T10:05:56+00:00",
      "cve": "CVE-2026-12570",
      "id": "CVE-2026-12570",
      "initial_release_date": "2026-08-10T06:29:56.111000+00:00",
      "product_status:known_affected": "9",
      "source": "Red Hat CSAF VEX",
      "status": "final",
      "title": "keras: Keras: Denial of Service via HDF5 Shape Bomb when loading malicious model files",
      "url": "https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-12570.json",
      "version": "3"
    },
    "vulnrichment": {
      "containers": {
        "adp": [
          {
            "metrics": [
              {
                "other": {
                  "content": {
                    "id": "CVE-2026-12570",
                    "options": [
                      {
                        "Exploitation": "poc"
                      },
                      {
                        "Automatable": "no"
                      },
                      {
                        "Technical Impact": "partial"
                      }
                    ],
                    "role": "CISA Coordinator",
                    "timestamp": "2026-08-10T17:37:17.637683Z",
                    "version": "2.0.3"
                  },
                  "type": "ssvc"
                }
              }
            ],
            "providerMetadata": {
              "dateUpdated": "2026-08-10T17:37:35.339Z",
              "orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
              "shortName": "CISA-ADP"
            },
            "references": [
              {
                "tags": [
                  "exploit"
                ],
                "url": "https://huntr.com/bounties/a064f475-780a-409a-82f7-678512f27ad8"
              }
            ],
            "title": "CISA ADP Vulnrichment"
          }
        ],
        "cna": {
          "affected": [
            {
              "product": "keras-team/keras",
              "vendor": "keras-team",
              "versions": [
                {
                  "lessThan": "3.12.3, 3.15.0",
                  "status": "affected",
                  "version": "unspecified",
                  "versionType": "custom"
                }
              ]
            }
          ],
          "descriptions": [
            {
              "lang": "en",
              "value": "A vulnerability in keras-team/keras versions \u003c= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function. The H5IOStore.__getitem__ method in keras/src/saving/saving_lib.py does not validate the shape or size of datasets, leading to unbounded memory allocation. A specially crafted .keras file can exploit this flaw to trigger an out-of-memory (OOM) condition, causing the process to be terminated (exit code 137). This issue bypasses the fix for CVE-2026-0897, which only addressed a similar vulnerability in KerasFileEditor. The attack vector includes poisoned models from public repositories or malicious model registries, posing a risk to machine learning pipelines that process untrusted models."
            }
          ],
          "metrics": [
            {
              "cvssV3_0": {
                "attackComplexity": "LOW",
                "attackVector": "LOCAL",
                "availabilityImpact": "HIGH",
                "baseScore": 5.5,
                "baseSeverity": "MEDIUM",
                "confidentialityImpact": "NONE",
                "integrityImpact": "NONE",
                "privilegesRequired": "NONE",
                "scope": "UNCHANGED",
                "userInteraction": "REQUIRED",
                "vectorString": "CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H",
                "version": "3.0"
              }
            }
          ],
          "problemTypes": [
            {
              "descriptions": [
                {
                  "cweId": "CWE-770",
                  "description": "CWE-770 Allocation of Resources Without Limits or Throttling",
                  "lang": "en",
                  "type": "CWE"
                }
              ]
            }
          ],
          "providerMetadata": {
            "dateUpdated": "2026-08-10T06:29:56.111Z",
            "orgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
            "shortName": "@huntr_ai"
          },
          "references": [
            {
              "url": "https://huntr.com/bounties/a064f475-780a-409a-82f7-678512f27ad8"
            },
            {
              "url": "https://github.com/keras-team/keras/commit/4933ea4a5b3fcc24ceacdc276f5bb5dfbd06756c"
            }
          ],
          "source": {
            "advisory": "a064f475-780a-409a-82f7-678512f27ad8",
            "discovery": "EXTERNAL"
          },
          "title": "Denial of Service via HDF5 Shape Bomb in keras.models.load_model() in keras-team/keras"
        }
      },
      "cveMetadata": {
        "assignerOrgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
        "assignerShortName": "@huntr_ai",
        "cveId": "CVE-2026-12570",
        "datePublished": "2026-08-10T06:29:56.111Z",
        "dateReserved": "2026-06-18T01:55:25.311Z",
        "dateUpdated": "2026-08-10T17:37:39.037Z",
        "state": "PUBLISHED"
      },
      "dataType": "CVE_RECORD",
      "dataVersion": "5.2"
    }
  }
}



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.