CVE-2026-1839 (GCVE-0-2026-1839)

Vulnerability from cvelistv5 – Published: 2026-04-07 05:22 – Updated: 2026-04-07 13:27
VLAI
Title
Arbitrary Code Execution via Unsafe torch.load() in Trainer Checkpoint Loading in huggingface/transformers
Summary
A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution. The `_load_rng_state()` method in `src/transformers/trainer.py` at line 3059 calls `torch.load()` without the `weights_only=True` parameter. This issue affects all versions of the library supporting `torch>=2.2` when used with PyTorch versions below 2.6, as the `safe_globals()` context manager provides no protection in these versions. An attacker can exploit this vulnerability by supplying a malicious checkpoint file, such as `rng_state.pth`, which can execute arbitrary code when loaded. The issue is resolved in version v5.0.0rc3.
Severity
6.5 (Medium)
CVSS:3.0/AV:L/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:H
SSVC
Exploitation: poc Automatable: no Technical Impact: total
CISA Coordinator · CISA-ADP (v2.0.3)
Decision recorded 2026-04-07 13:27 UTC
CWE
  • CWE-502 - Deserialization of Untrusted Data
Assigner
References
Impacted products
Vendor Product Version CPE status
huggingface huggingface/transformers Affected: unspecified , < v5.0.0rc3 (custom)
guessed Create a notification for this product.
Show details on NVD website
To clipboard 

{
  "containers": {
    "adp": [
      {
        "metrics": [
          {
            "other": {
              "content": {
                "id": "CVE-2026-1839",
                "options": [
                  {
                    "Exploitation": "poc"
                  },
                  {
                    "Automatable": "no"
                  },
                  {
                    "Technical Impact": "total"
                  }
                ],
                "role": "CISA Coordinator",
                "timestamp": "2026-04-07T13:27:38.099805Z",
                "version": "2.0.3"
              },
              "type": "ssvc"
            }
          }
        ],
        "providerMetadata": {
          "dateUpdated": "2026-04-07T13:27:41.789Z",
          "orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
          "shortName": "CISA-ADP"
        },
        "references": [
          {
            "tags": [
              "exploit"
            ],
            "url": "https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485"
          }
        ],
        "title": "CISA ADP Vulnrichment"
      }
    ],
    "cna": {
      "affected": [
        {
          "product": "huggingface/transformers",
          "vendor": "huggingface",
          "versions": [
            {
              "lessThan": "v5.0.0rc3",
              "status": "affected",
              "version": "unspecified",
              "versionType": "custom"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution. The `_load_rng_state()` method in `src/transformers/trainer.py` at line 3059 calls `torch.load()` without the `weights_only=True` parameter. This issue affects all versions of the library supporting `torch\u003e=2.2` when used with PyTorch versions below 2.6, as the `safe_globals()` context manager provides no protection in these versions. An attacker can exploit this vulnerability by supplying a malicious checkpoint file, such as `rng_state.pth`, which can execute arbitrary code when loaded. The issue is resolved in version v5.0.0rc3."
        }
      ],
      "metrics": [
        {
          "cvssV3_0": {
            "attackComplexity": "HIGH",
            "attackVector": "LOCAL",
            "availabilityImpact": "HIGH",
            "baseScore": 6.5,
            "baseSeverity": "MEDIUM",
            "confidentialityImpact": "HIGH",
            "integrityImpact": "LOW",
            "privilegesRequired": "NONE",
            "scope": "UNCHANGED",
            "userInteraction": "REQUIRED",
            "vectorString": "CVSS:3.0/AV:L/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:H",
            "version": "3.0"
          }
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "cweId": "CWE-502",
              "description": "CWE-502 Deserialization of Untrusted Data",
              "lang": "en",
              "type": "CWE"
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-04-07T05:22:00.959Z",
        "orgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
        "shortName": "@huntr_ai"
      },
      "references": [
        {
          "url": "https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485"
        },
        {
          "url": "https://github.com/huggingface/transformers/commit/03c8082ba4594c9b8d6fe190ca9bed0e5f8ca396"
        }
      ],
      "source": {
        "advisory": "3c77bb97-e493-493d-9a88-c57f5c536485",
        "discovery": "EXTERNAL"
      },
      "title": "Arbitrary Code Execution via Unsafe torch.load() in Trainer Checkpoint Loading in huggingface/transformers"
    }
  },
  "cveMetadata": {
    "assignerOrgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
    "assignerShortName": "@huntr_ai",
    "cveId": "CVE-2026-1839",
    "datePublished": "2026-04-07T05:22:00.959Z",
    "dateReserved": "2026-02-03T16:49:27.781Z",
    "dateUpdated": "2026-04-07T13:27:41.789Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2",
  "vulnerability-lookup:meta": {
    "epss": {
      "cve": "CVE-2026-1839",
      "date": "2026-09-20",
      "epss": "0.00349",
      "percentile": "0.28546"
    },
    "nvd": {
      "cve": {
        "affected": [
          {
            "affectedData": [
              {
                "product": "huggingface/transformers",
                "vendor": "huggingface",
                "versions": [
                  {
                    "lessThan": "v5.0.0rc3",
                    "status": "affected",
                    "version": "unspecified",
                    "versionType": "custom"
                  }
                ]
              }
            ],
            "source": "security@huntr.dev"
          }
        ],
        "configurations": [
          {
            "nodes": [
              {
                "cpeMatch": [
                  {
                    "criteria": "cpe:2.3:a:huggingface:transformers:*:*:*:*:*:*:*:*",
                    "matchCriteriaId": "501E3796-D201-494C-B59D-FA598DC36738",
                    "versionEndExcluding": "5.0.0",
                    "vulnerable": true
                  },
                  {
                    "criteria": "cpe:2.3:a:huggingface:transformers:5.0.0:rc0:*:*:*:*:*:*",
                    "matchCriteriaId": "70A12094-61B0-49E7-9663-73E5B874E819",
                    "vulnerable": true
                  },
                  {
                    "criteria": "cpe:2.3:a:huggingface:transformers:5.0.0:rc1:*:*:*:*:*:*",
                    "matchCriteriaId": "424574B4-B321-4D5A-BE48-95A61D9F3752",
                    "vulnerable": true
                  },
                  {
                    "criteria": "cpe:2.3:a:huggingface:transformers:5.0.0:rc2:*:*:*:*:*:*",
                    "matchCriteriaId": "B3F5BCCD-1960-4809-9519-4ADDACAFB3D3",
                    "vulnerable": true
                  }
                ],
                "negate": false,
                "operator": "OR"
              }
            ]
          }
        ],
        "cveTags": [],
        "descriptions": [
          {
            "lang": "en",
            "value": "A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution. The `_load_rng_state()` method in `src/transformers/trainer.py` at line 3059 calls `torch.load()` without the `weights_only=True` parameter. This issue affects all versions of the library supporting `torch\u003e=2.2` when used with PyTorch versions below 2.6, as the `safe_globals()` context manager provides no protection in these versions. An attacker can exploit this vulnerability by supplying a malicious checkpoint file, such as `rng_state.pth`, which can execute arbitrary code when loaded. The issue is resolved in version v5.0.0rc3."
          }
        ],
        "id": "CVE-2026-1839",
        "lastModified": "2026-06-17T10:16:35.823",
        "metrics": {
          "cvssMetricV30": [
            {
              "cvssData": {
                "attackComplexity": "HIGH",
                "attackVector": "LOCAL",
                "availabilityImpact": "HIGH",
                "baseScore": 6.5,
                "baseSeverity": "MEDIUM",
                "confidentialityImpact": "HIGH",
                "integrityImpact": "LOW",
                "privilegesRequired": "NONE",
                "scope": "UNCHANGED",
                "userInteraction": "REQUIRED",
                "vectorString": "CVSS:3.0/AV:L/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:H",
                "version": "3.0"
              },
              "exploitabilityScore": 1.0,
              "impactScore": 5.5,
              "source": "security@huntr.dev",
              "type": "Secondary"
            }
          ],
          "cvssMetricV31": [
            {
              "cvssData": {
                "attackComplexity": "LOW",
                "attackVector": "LOCAL",
                "availabilityImpact": "HIGH",
                "baseScore": 7.8,
                "baseSeverity": "HIGH",
                "confidentialityImpact": "HIGH",
                "integrityImpact": "HIGH",
                "privilegesRequired": "NONE",
                "scope": "UNCHANGED",
                "userInteraction": "REQUIRED",
                "vectorString": "CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H",
                "version": "3.1"
              },
              "exploitabilityScore": 1.8,
              "impactScore": 5.9,
              "source": "nvd@nist.gov",
              "type": "Primary"
            }
          ],
          "ssvcV203": [
            {
              "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
              "ssvcData": {
                "id": "CVE-2026-1839",
                "options": [
                  {
                    "exploitation": "poc"
                  },
                  {
                    "automatable": "no"
                  },
                  {
                    "technicalImpact": "total"
                  }
                ],
                "role": "CISA Coordinator",
                "timestamp": "2026-04-07T13:27:38.099805Z",
                "version": "2.0.3"
              }
            }
          ]
        },
        "published": "2026-04-07T06:16:41.490",
        "references": [
          {
            "source": "security@huntr.dev",
            "tags": [
              "Patch"
            ],
            "url": "https://github.com/huggingface/transformers/commit/03c8082ba4594c9b8d6fe190ca9bed0e5f8ca396"
          },
          {
            "source": "security@huntr.dev",
            "tags": [
              "Exploit",
              "Third Party Advisory"
            ],
            "url": "https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485"
          },
          {
            "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
            "tags": [
              "Exploit",
              "Third Party Advisory"
            ],
            "url": "https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485"
          }
        ],
        "sourceIdentifier": "security@huntr.dev",
        "vulnStatus": "Analyzed",
        "weaknesses": [
          {
            "description": [
              {
                "lang": "en",
                "value": "CWE-502"
              }
            ],
            "source": "security@huntr.dev",
            "type": "Secondary"
          }
        ]
      }
    },
    "redhat_vex": {
      "aggregate_severity": "Moderate",
      "current_release_date": "2026-06-30T11:45:21+00:00",
      "cve": "CVE-2026-1839",
      "id": "CVE-2026-1839",
      "initial_release_date": "2026-04-07T05:22:00.959000+00:00",
      "product_status:known_affected": "53",
      "source": "Red Hat CSAF VEX",
      "status": "final",
      "title": "transformers: HuggingFace Transformers: Arbitrary code execution via malicious checkpoint file",
      "url": "https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-1839.json",
      "version": "3"
    },
    "vulnrichment": {
      "containers": {
        "adp": [
          {
            "metrics": [
              {
                "other": {
                  "content": {
                    "id": "CVE-2026-1839",
                    "options": [
                      {
                        "Exploitation": "poc"
                      },
                      {
                        "Automatable": "no"
                      },
                      {
                        "Technical Impact": "total"
                      }
                    ],
                    "role": "CISA Coordinator",
                    "timestamp": "2026-04-07T13:27:38.099805Z",
                    "version": "2.0.3"
                  },
                  "type": "ssvc"
                }
              }
            ],
            "providerMetadata": {
              "dateUpdated": "2026-04-07T13:27:31.816Z",
              "orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
              "shortName": "CISA-ADP"
            },
            "references": [
              {
                "tags": [
                  "exploit"
                ],
                "url": "https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485"
              }
            ],
            "title": "CISA ADP Vulnrichment"
          }
        ],
        "cna": {
          "affected": [
            {
              "product": "huggingface/transformers",
              "vendor": "huggingface",
              "versions": [
                {
                  "lessThan": "v5.0.0rc3",
                  "status": "affected",
                  "version": "unspecified",
                  "versionType": "custom"
                }
              ]
            }
          ],
          "descriptions": [
            {
              "lang": "en",
              "value": "A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution. The `_load_rng_state()` method in `src/transformers/trainer.py` at line 3059 calls `torch.load()` without the `weights_only=True` parameter. This issue affects all versions of the library supporting `torch\u003e=2.2` when used with PyTorch versions below 2.6, as the `safe_globals()` context manager provides no protection in these versions. An attacker can exploit this vulnerability by supplying a malicious checkpoint file, such as `rng_state.pth`, which can execute arbitrary code when loaded. The issue is resolved in version v5.0.0rc3."
            }
          ],
          "metrics": [
            {
              "cvssV3_0": {
                "attackComplexity": "HIGH",
                "attackVector": "LOCAL",
                "availabilityImpact": "HIGH",
                "baseScore": 6.5,
                "baseSeverity": "MEDIUM",
                "confidentialityImpact": "HIGH",
                "integrityImpact": "LOW",
                "privilegesRequired": "NONE",
                "scope": "UNCHANGED",
                "userInteraction": "REQUIRED",
                "vectorString": "CVSS:3.0/AV:L/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:H",
                "version": "3.0"
              }
            }
          ],
          "problemTypes": [
            {
              "descriptions": [
                {
                  "cweId": "CWE-502",
                  "description": "CWE-502 Deserialization of Untrusted Data",
                  "lang": "en",
                  "type": "CWE"
                }
              ]
            }
          ],
          "providerMetadata": {
            "dateUpdated": "2026-04-07T05:22:00.959Z",
            "orgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
            "shortName": "@huntr_ai"
          },
          "references": [
            {
              "url": "https://huntr.com/bounties/3c77bb97-e493-493d-9a88-c57f5c536485"
            },
            {
              "url": "https://github.com/huggingface/transformers/commit/03c8082ba4594c9b8d6fe190ca9bed0e5f8ca396"
            }
          ],
          "source": {
            "advisory": "3c77bb97-e493-493d-9a88-c57f5c536485",
            "discovery": "EXTERNAL"
          },
          "title": "Arbitrary Code Execution via Unsafe torch.load() in Trainer Checkpoint Loading in huggingface/transformers"
        }
      },
      "cveMetadata": {
        "assignerOrgId": "c09c270a-b464-47c1-9133-acb35b22c19a",
        "assignerShortName": "@huntr_ai",
        "cveId": "CVE-2026-1839",
        "datePublished": "2026-04-07T05:22:00.959Z",
        "dateReserved": "2026-02-03T16:49:27.781Z",
        "dateUpdated": "2026-04-07T13:27:41.789Z",
        "state": "PUBLISHED"
      },
      "dataType": "CVE_RECORD",
      "dataVersion": "5.2"
    }
  }
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





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.


Detection rules are retrieved from Rulezet.


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.