CVE-2026-12481

{"dataType": "CVE_RECORD", "dataVersion": "5.2", "cveMetadata": {"cveId": "CVE-2026-12481", "assignerOrgId": "c09c270a-b464-47c1-9133-acb35b22c19a", "state": "PUBLISHED", "assignerShortName": "@huntr_ai", "dateReserved": "2026-06-17T01:06:48.759Z", "datePublished": "2026-07-03T20:36:05.003Z", "dateUpdated": "2026-07-06T15:15:46.244Z"}, "containers": {"cna": {"title": "Deserialization of Untrusted Data in keras-team/keras", "providerMetadata": {"orgId": "c09c270a-b464-47c1-9133-acb35b22c19a", "shortName": "@huntr_ai", "dateUpdated": "2026-07-03T20:36:05.003Z"}, "descriptions": [{"lang": "en", "value": "A vulnerability in keras-team/keras version 3.14.0 allows for arbitrary code execution due to improper handling of deserialization in the `Lambda` layer. Specifically, the `_raise_for_lambda_deserialization()` function fails to enforce the safe-mode guard when `safe_mode` is set to `None`, which is the default value when `from_config()` is called outside of a `SafeModeScope` context. This logic error conflates `None` (unset/default-deny) with `False` (explicitly disabled), bypassing the guard and allowing attacker-controlled `marshal` bytecode to be deserialized. Affected call sites include `keras.layers.deserialize(config)`, `keras.models.clone_model(model)`, and any direct invocation of `Lambda.from_config(config)` without an enclosing `SafeModeScope(True)`. This vulnerability can be exploited to achieve arbitrary OS-level code execution in the context of the server or user process."}], "affected": [{"vendor": "keras-team", "product": "keras-team/keras", "versions": [{"version": "unspecified", "status": "affected", "versionType": "custom", "lessThanOrEqual": "latest"}]}], "references": [{"url": "https://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d"}], "metrics": [{"cvssV3_0": {"version": "3.0", "attackComplexity": "LOW", "attackVector": "NETWORK", "availabilityImpact": "HIGH", "confidentialityImpact": "HIGH", "integrityImpact": "HIGH", "privilegesRequired": "NONE", "scope": "UNCHANGED", "userInteraction": "REQUIRED", "vectorString": "CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H", "baseScore": 8.8, "baseSeverity": "HIGH"}}], "problemTypes": [{"descriptions": [{"type": "CWE", "lang": "en", "description": "CWE-502 Deserialization of Untrusted Data", "cweId": "CWE-502"}]}], "source": {"advisory": "59ceaed1-c8a3-4135-8f94-169ade02823d", "discovery": "EXTERNAL"}}, "adp": [{"references": [{"url": "https://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d", "tags": ["exploit"]}], "metrics": [{"other": {"type": "ssvc", "content": {"timestamp": "2026-07-06T15:15:34.297007Z", "id": "CVE-2026-12481", "options": [{"Exploitation": "poc"}, {"Automatable": "no"}, {"Technical Impact": "total"}], "role": "CISA Coordinator", "version": "2.0.3"}}}], "title": "CISA ADP Vulnrichment", "providerMetadata": {"orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "shortName": "CISA-ADP", "dateUpdated": "2026-07-06T15:15:46.244Z"}}]}}

{"dataType": "CVE_RECORD", "containers": {"adp": [{"title": "CISA ADP Vulnrichment", "metrics": [{"other": {"type": "ssvc", "content": {"id": "CVE-2026-12481", "role": "CISA Coordinator", "options": [{"Exploitation": "poc"}, {"Automatable": "no"}, {"Technical Impact": "total"}], "version": "2.0.3", "timestamp": "2026-07-06T15:15:34.297007Z"}}}], "references": [{"url": "https://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d", "tags": ["exploit"]}], "providerMetadata": {"orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "shortName": "CISA-ADP", "dateUpdated": "2026-07-06T15:15:28.773Z"}}], "cna": {"title": "Deserialization of Untrusted Data in keras-team/keras", "source": {"advisory": "59ceaed1-c8a3-4135-8f94-169ade02823d", "discovery": "EXTERNAL"}, "metrics": [{"cvssV3_0": {"scope": "UNCHANGED", "version": "3.0", "baseScore": 8.8, "attackVector": "NETWORK", "baseSeverity": "HIGH", "vectorString": "CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H", "integrityImpact": "HIGH", "userInteraction": "REQUIRED", "attackComplexity": "LOW", "availabilityImpact": "HIGH", "privilegesRequired": "NONE", "confidentialityImpact": "HIGH"}}], "affected": [{"vendor": "keras-team", "product": "keras-team/keras", "versions": [{"status": "affected", "version": "unspecified", "versionType": "custom", "lessThanOrEqual": "latest"}]}], "references": [{"url": "https://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d"}], "descriptions": [{"lang": "en", "value": "A vulnerability in keras-team/keras version 3.14.0 allows for arbitrary code execution due to improper handling of deserialization in the `Lambda` layer. Specifically, the `_raise_for_lambda_deserialization()` function fails to enforce the safe-mode guard when `safe_mode` is set to `None`, which is the default value when `from_config()` is called outside of a `SafeModeScope` context. This logic error conflates `None` (unset/default-deny) with `False` (explicitly disabled), bypassing the guard and allowing attacker-controlled `marshal` bytecode to be deserialized. Affected call sites include `keras.layers.deserialize(config)`, `keras.models.clone_model(model)`, and any direct invocation of `Lambda.from_config(config)` without an enclosing `SafeModeScope(True)`. This vulnerability can be exploited to achieve arbitrary OS-level code execution in the context of the server or user process."}], "problemTypes": [{"descriptions": [{"lang": "en", "type": "CWE", "cweId": "CWE-502", "description": "CWE-502 Deserialization of Untrusted Data"}]}], "providerMetadata": {"orgId": "c09c270a-b464-47c1-9133-acb35b22c19a", "shortName": "@huntr_ai", "dateUpdated": "2026-07-03T20:36:05.003Z"}}}, "cveMetadata": {"cveId": "CVE-2026-12481", "state": "PUBLISHED", "dateUpdated": "2026-07-06T15:15:46.244Z", "dateReserved": "2026-06-17T01:06:48.759Z", "assignerOrgId": "c09c270a-b464-47c1-9133-acb35b22c19a", "datePublished": "2026-07-03T20:36:05.003Z", "assignerShortName": "@huntr_ai"}, "dataVersion": "5.2"}

{"affected": [{"affectedData": [{"product": "keras-team/keras", "vendor": "keras-team", "versions": [{"lessThanOrEqual": "latest", "status": "affected", "version": "unspecified", "versionType": "custom"}]}], "source": "security@huntr.dev"}], "configurations": [{"nodes": [{"cpeMatch": [{"criteria": "cpe:2.3:a:keras:keras:3.14.0:*:*:*:*:*:*:*", "matchCriteriaId": "5AB5B43D-0D41-4CBB-97B5-F5FE10D56689", "vulnerable": true}], "negate": false, "operator": "OR"}]}], "cveTags": [], "descriptions": [{"lang": "en", "value": "A vulnerability in keras-team/keras version 3.14.0 allows for arbitrary code execution due to improper handling of deserialization in the `Lambda` layer. Specifically, the `_raise_for_lambda_deserialization()` function fails to enforce the safe-mode guard when `safe_mode` is set to `None`, which is the default value when `from_config()` is called outside of a `SafeModeScope` context. This logic error conflates `None` (unset/default-deny) with `False` (explicitly disabled), bypassing the guard and allowing attacker-controlled `marshal` bytecode to be deserialized. Affected call sites include `keras.layers.deserialize(config)`, `keras.models.clone_model(model)`, and any direct invocation of `Lambda.from_config(config)` without an enclosing `SafeModeScope(True)`. This vulnerability can be exploited to achieve arbitrary OS-level code execution in the context of the server or user process."}, {"lang": "es", "value": "Una vulnerabilidad en keras-team/keras versi\u00f3n 3.14.0 permite la ejecuci\u00f3n de c\u00f3digo arbitrario debido a un manejo inadecuado de la deserializaci\u00f3n en la capa \u2019Lambda\u2019. Espec\u00edficamente, la funci\u00f3n \u2019_raise_for_lambda_deserialization\u2019 no aplica la protecci\u00f3n de modo seguro cuando \u2019safe_mode\u2019 se establece en \u2019None\u2019, que es el valor predeterminado cuando se llama a \u2019from_config()\u2019 fuera de un contexto de \u2019SafeModeScope\u2019. Este error l\u00f3gico confunde \u2019None\u2019 (no establecido/denegaci\u00f3n predeterminada) con \u2019False\u2019 (expl\u00edcitamente deshabilitado), eludiendo la protecci\u00f3n y permitiendo que el bytecode \u2019marshal\u2019 controlado por el atacante sea deserializado. Los sitios de llamada afectados incluyen \u2019keras.layers.deserialize(config)\u2019, \u2019keras.models.clone_model(model)\u2019, y cualquier invocaci\u00f3n directa de \u2019Lambda.from_config(config)\u2019 sin un \u2019SafeModeScope(True)\u2019 envolvente. Esta vulnerabilidad puede ser explotada para lograr la ejecuci\u00f3n de c\u00f3digo arbitrario a nivel del sistema operativo en el contexto del servidor o del proceso de usuario."}], "id": "CVE-2026-12481", "lastModified": "2026-10-25T15:10:00.147", "metrics": {"cvssMetricV30": [{"cvssData": {"attackComplexity": "LOW", "attackVector": "NETWORK", "availabilityImpact": "HIGH", "baseScore": 8.8, "baseSeverity": "HIGH", "confidentialityImpact": "HIGH", "integrityImpact": "HIGH", "privilegesRequired": "NONE", "scope": "UNCHANGED", "userInteraction": "REQUIRED", "vectorString": "CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H", "version": "3.0"}, "exploitabilityScore": 2.8, "impactScore": 5.9, "source": "security@huntr.dev", "type": "Secondary"}], "cvssMetricV31": [{"cvssData": {"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"}, "exploitabilityScore": 3.9, "impactScore": 5.9, "source": "nvd@nist.gov", "type": "Primary"}], "ssvcV203": [{"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "ssvcData": {"id": "CVE-2026-12481", "options": [{"exploitation": "poc"}, {"automatable": "no"}, {"technicalImpact": "total"}], "role": "CISA Coordinator", "timestamp": "2026-07-06T15:15:34.297007Z", "version": "2.0.3"}}]}, "published": "2026-07-03T21:16:54.737", "references": [{"source": "security@huntr.dev", "tags": ["Exploit", "Third Party Advisory"], "url": "https://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d"}, {"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "tags": ["Exploit", "Third Party Advisory"], "url": "https://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d"}], "sourceIdentifier": "security@huntr.dev", "vulnStatus": "Analyzed", "weaknesses": [{"description": [{"lang": "en", "value": "CWE-502"}], "source": "security@huntr.dev", "type": "Secondary"}]}

{"bugzilla": {"description": "keras: Keras: Arbitrary code execution via deserialization vulnerability", "id": "2496942", "url": "https://bugzilla.redhat.com/show_bug.cgi?id=2496942"}, "csaw": false, "cvss3": {"cvss3_base_score": "8.8", "cvss3_scoring_vector": "CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H", "status": "draft"}, "cwe": "CWE-502", "details": ["A vulnerability in keras-team/keras version 3.14.0 allows for arbitrary code execution due to improper handling of deserialization in the `Lambda` layer. Specifically, the `_raise_for_lambda_deserialization()` function fails to enforce the safe-mode guard when `safe_mode` is set to `None`, which is the default value when `from_config()` is called outside of a `SafeModeScope` context. This logic error conflates `None` (unset/default-deny) with `False` (explicitly disabled), bypassing the guard and allowing attacker-controlled `marshal` bytecode to be deserialized. Affected call sites include `keras.layers.deserialize(config)`, `keras.models.clone_model(model)`, and any direct invocation of `Lambda.from_config(config)` without an enclosing `SafeModeScope(True)`. This vulnerability can be exploited to achieve arbitrary OS-level code execution in the context of the server or user process.", "A flaw was found in the Keras deep learning library. This vulnerability allows a remote attacker to execute arbitrary code on the system by exploiting improper handling of deserialization in the `Lambda` layer. Specifically, a security safeguard designed to prevent unsafe deserialization is bypassed when the `safe_mode` setting is not explicitly enabled, allowing malicious code to be processed. This can lead to complete compromise of the affected server or user process."], "mitigation": {"lang": "en:us", "value": "To reduce exposure, only deserialize Keras models from trusted sources. When deserializing models, explicitly enable `safe_mode` by wrapping the deserialization call within a `SafeModeScope(True)` context. This ensures the deserialization safeguard is enforced, preventing the execution of arbitrary code."}, "name": "CVE-2026-12481", "package_state": [{"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Not affected", "package_name": "rhoai/odh-kserve-agent-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Not affected", "package_name": "rhoai/odh-kserve-controller-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Not affected", "package_name": "rhoai/odh-kserve-router-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Affected", "package_name": "rhoai/odh-kserve-storage-initializer-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Affected", "package_name": "rhoai/odh-modelmesh-runtime-adapter-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Affected", "package_name": "rhoai/odh-pipeline-runtime-tensorflow-cuda-py312-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Affected", "package_name": "rhoai/odh-pipeline-runtime-tensorflow-rocm-py312-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Affected", "package_name": "rhoai/odh-workbench-jupyter-tensorflow-cuda-py312-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}, {"cpe": "cpe:/a:redhat:openshift_ai", "fix_state": "Affected", "package_name": "rhoai/odh-workbench-jupyter-tensorflow-rocm-py312-rhel9", "product_name": "Red Hat OpenShift AI (RHOAI)"}], "public_date": "2026-07-03T20:36:05Z", "references": ["https://www.cve.org/CVERecord?id=CVE-2026-12481\nhttps://nvd.nist.gov/vuln/detail/CVE-2026-12481\nhttps://huntr.com/bounties/59ceaed1-c8a3-4135-8f94-169ade02823d"], "statement": "This is an Important arbitrary code execution vulnerability in the Keras deep learning library, impacting Red Hat OpenShift AI components. The flaw arises from improper deserialization in the `Lambda` layer, where a security safeguard is bypassed if `safe_mode` is not explicitly enabled. Exploitation requires user interaction, such as loading a specially crafted Keras model.", "threat_severity": "Important"}

{"affected": [{"configurations": [{"platform": null, "status": "affected", "versions": {"scheme": "generic", "value": "[0,*]"}}], "enrichment": {"confidence": 100.0, "confidence_source": "manual", "scores": [{"score": 100.0, "source": "manual"}]}, "original": {"product": "keras-team/keras", "source": "cna", "vendor": "keras-team"}, "product": "keras", "vendor": "keras-team"}], "created": "2026-07-04T08:45:02.985516+00:00", "updated": "2026-08-12T00:45:03.305186+00:00", "vendors": ["keras-team", "keras-team$PRODUCT$keras"]}