CVE-2026-31253 (GCVE-0-2026-31253)
Vulnerability from cvelistv5 – Published: 2026-05-11 00:00 – Updated: 2026-05-12 19:32
VLAI
EPSS
VEX
Summary
The flash-attention training framework thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains an insecure deserialization vulnerability (CWE-502) in its checkpoint loading mechanism. The load_checkpoint() function in checkpoint.py and the checkpoint loading code in eval.py use torch.load() 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 providing a maliciously crafted checkpoint file. When a victim loads this checkpoint during model warmstarting or evaluation, arbitrary code is executed on the victim's system.
Severity
7.3 (High)
SSVC
Exploitation: none
Automatable: no
Technical Impact: partial
CISA Coordinator · CISA-ADP (v2.0.3)
Decision recorded 2026-05-12 19:32 UTC
CWE
Assigner
References
{
"containers": {
"adp": [
{
"metrics": [
{
"cvssV3_1": {
"attackComplexity": "LOW",
"attackVector": "NETWORK",
"availabilityImpact": "LOW",
"baseScore": 7.3,
"baseSeverity": "HIGH",
"confidentialityImpact": "LOW",
"integrityImpact": "LOW",
"privilegesRequired": "NONE",
"scope": "UNCHANGED",
"userInteraction": "NONE",
"vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L",
"version": "3.1"
}
},
{
"other": {
"content": {
"id": "CVE-2026-31253",
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "no"
},
{
"Technical Impact": "partial"
}
],
"role": "CISA Coordinator",
"timestamp": "2026-05-12T19:32:06.944533Z",
"version": "2.0.3"
},
"type": "ssvc"
}
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-502",
"description": "CWE-502 Deserialization of Untrusted Data",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-94",
"description": "CWE-94 Improper Control of Generation of Code (\u0027Code Injection\u0027)",
"lang": "en",
"type": "CWE"
}
]
}
],
"providerMetadata": {
"dateUpdated": "2026-05-12T19:32:41.020Z",
"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 flash-attention training framework thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains an insecure deserialization vulnerability (CWE-502) in its checkpoint loading mechanism. The load_checkpoint() function in checkpoint.py and the checkpoint loading code in eval.py use torch.load() 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 providing a maliciously crafted checkpoint file. When a victim loads this checkpoint during model warmstarting or evaluation, arbitrary code is executed on the victim\u0027s system."
}
],
"problemTypes": [
{
"descriptions": [
{
"description": "n/a",
"lang": "en",
"type": "text"
}
]
}
],
"providerMetadata": {
"dateUpdated": "2026-05-11T15:40:10.644Z",
"orgId": "8254265b-2729-46b6-b9e3-3dfca2d5bfca",
"shortName": "mitre"
},
"references": [
{
"url": "https://github.com/Dao-AILab/flash-attention"
},
{
"url": "https://www.notion.so/CVE-2026-31253-35d1e1393188813f9e77e2038104bc49"
}
]
}
},
"cveMetadata": {
"assignerOrgId": "8254265b-2729-46b6-b9e3-3dfca2d5bfca",
"assignerShortName": "mitre",
"cveId": "CVE-2026-31253",
"datePublished": "2026-05-11T00:00:00.000Z",
"dateReserved": "2026-03-09T00:00:00.000Z",
"dateUpdated": "2026-05-12T19:32:41.020Z",
"state": "PUBLISHED"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2",
"vulnerability-lookup:meta": {
"epss": {
"cve": "CVE-2026-31253",
"date": "2026-09-20",
"epss": "0.00218",
"percentile": "0.12563"
},
"nvd": {
"cve": {
"affected": [
{
"affectedData": [
{
"product": "n/a",
"vendor": "n/a",
"versions": [
{
"status": "affected",
"version": "n/a"
}
]
}
],
"source": "cve@mitre.org"
}
],
"cveTags": [],
"descriptions": [
{
"lang": "en",
"value": "The flash-attention training framework thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains an insecure deserialization vulnerability (CWE-502) in its checkpoint loading mechanism. The load_checkpoint() function in checkpoint.py and the checkpoint loading code in eval.py use torch.load() 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 providing a maliciously crafted checkpoint file. When a victim loads this checkpoint during model warmstarting or evaluation, arbitrary code is executed on the victim\u0027s system."
}
],
"id": "CVE-2026-31253",
"lastModified": "2026-06-17T10:33:32.820",
"metrics": {
"cvssMetricV31": [
{
"cvssData": {
"attackComplexity": "LOW",
"attackVector": "NETWORK",
"availabilityImpact": "LOW",
"baseScore": 7.3,
"baseSeverity": "HIGH",
"confidentialityImpact": "LOW",
"integrityImpact": "LOW",
"privilegesRequired": "NONE",
"scope": "UNCHANGED",
"userInteraction": "NONE",
"vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L",
"version": "3.1"
},
"exploitabilityScore": 3.9,
"impactScore": 3.4,
"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"type": "Secondary"
}
],
"ssvcV203": [
{
"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"ssvcData": {
"id": "CVE-2026-31253",
"options": [
{
"exploitation": "none"
},
{
"automatable": "no"
},
{
"technicalImpact": "partial"
}
],
"role": "CISA Coordinator",
"timestamp": "2026-05-12T19:32:06.944533Z",
"version": "2.0.3"
}
}
]
},
"published": "2026-05-11T17:16:20.307",
"references": [
{
"source": "cve@mitre.org",
"url": "https://github.com/Dao-AILab/flash-attention"
},
{
"source": "cve@mitre.org",
"url": "https://www.notion.so/CVE-2026-31253-35d1e1393188813f9e77e2038104bc49"
}
],
"sourceIdentifier": "cve@mitre.org",
"vulnStatus": "Deferred",
"weaknesses": [
{
"description": [
{
"lang": "en",
"value": "CWE-94"
},
{
"lang": "en",
"value": "CWE-502"
}
],
"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"type": "Secondary"
}
]
}
},
"redhat_vex": {
"current_release_date": "2026-05-12T20:21:51+00:00",
"cve": "CVE-2026-31253",
"id": "CVE-2026-31253",
"initial_release_date": "2026-01-01T00:00:00+00:00",
"product_status:known_not_affected": "1",
"source": "Red Hat CSAF VEX",
"status": "final",
"title": "CVE-2026-31253",
"url": "https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-31253.json",
"version": "3"
},
"vulnrichment": {
"containers": {
"adp": [
{
"metrics": [
{
"cvssV3_1": {
"attackComplexity": "LOW",
"attackVector": "NETWORK",
"availabilityImpact": "LOW",
"baseScore": 7.3,
"baseSeverity": "HIGH",
"confidentialityImpact": "LOW",
"integrityImpact": "LOW",
"privilegesRequired": "NONE",
"scope": "UNCHANGED",
"userInteraction": "NONE",
"vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L",
"version": "3.1"
}
},
{
"other": {
"content": {
"id": "CVE-2026-31253",
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "no"
},
{
"Technical Impact": "partial"
}
],
"role": "CISA Coordinator",
"timestamp": "2026-05-12T19:32:06.944533Z",
"version": "2.0.3"
},
"type": "ssvc"
}
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-502",
"description": "CWE-502 Deserialization of Untrusted Data",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-94",
"description": "CWE-94 Improper Control of Generation of Code (\u0027Code Injection\u0027)",
"lang": "en",
"type": "CWE"
}
]
}
],
"providerMetadata": {
"dateUpdated": "2026-05-12T19:32:36.406Z",
"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 flash-attention training framework thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains an insecure deserialization vulnerability (CWE-502) in its checkpoint loading mechanism. The load_checkpoint() function in checkpoint.py and the checkpoint loading code in eval.py use torch.load() 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 providing a maliciously crafted checkpoint file. When a victim loads this checkpoint during model warmstarting or evaluation, arbitrary code is executed on the victim\u0027s system."
}
],
"problemTypes": [
{
"descriptions": [
{
"description": "n/a",
"lang": "en",
"type": "text"
}
]
}
],
"providerMetadata": {
"dateUpdated": "2026-05-11T15:40:10.644Z",
"orgId": "8254265b-2729-46b6-b9e3-3dfca2d5bfca",
"shortName": "mitre"
},
"references": [
{
"url": "https://github.com/Dao-AILab/flash-attention"
},
{
"url": "https://www.notion.so/CVE-2026-31253-35d1e1393188813f9e77e2038104bc49"
}
]
}
},
"cveMetadata": {
"assignerOrgId": "8254265b-2729-46b6-b9e3-3dfca2d5bfca",
"assignerShortName": "mitre",
"cveId": "CVE-2026-31253",
"datePublished": "2026-05-11T00:00:00.000Z",
"dateReserved": "2026-03-09T00:00:00.000Z",
"dateUpdated": "2026-05-11T15:40:10.644Z",
"state": "PUBLISHED"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2"
}
}
}
Loading…
Loading…
Experimental. This forecast is provided for visualization only and may change without notice. Do not use it for operational decisions.
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.
Loading…
The MITRE ATT&CK techniques below are AI-generated suggestions, inferred from the description of the
vulnerability by the CIRCL/vulnerability-attack-technique-classification-roberta-base
model, served locally by ML-Gateway.
They have not been verified by an analyst and are provided for guidance only.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Loading…
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.
Loading…