CVE-2026-14535

{"id": "CVE-2026-14535", "cveTags": [], "metrics": {"ssvcV203": [{"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "ssvcData": {"id": "CVE-2026-14535", "role": "CISA Coordinator", "options": [{"exploitation": "poc"}, {"automatable": "no"}, {"technicalImpact": "total"}], "version": "2.0.3", "timestamp": "2026-07-06T14:58:30.612222Z"}}], "cvssMetricV31": [{"type": "Secondary", "source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082", "cvssData": {"scope": "UNCHANGED", "version": "3.1", "baseScore": 8.8, "attackVector": "NETWORK", "baseSeverity": "HIGH", "vectorString": "CVSS:3.1/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"}, "impactScore": 5.9, "exploitabilityScore": 2.8}, {"type": "Primary", "source": "nvd@nist.gov", "cvssData": {"scope": "UNCHANGED", "version": "3.1", "baseScore": 9.8, "attackVector": "NETWORK", "baseSeverity": "CRITICAL", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H", "integrityImpact": "HIGH", "userInteraction": "NONE", "attackComplexity": "LOW", "availabilityImpact": "HIGH", "privilegesRequired": "NONE", "confidentialityImpact": "HIGH"}, "impactScore": 5.9, "exploitabilityScore": 3.9}]}, "affected": [{"source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082", "affectedData": [{"vendor": "trailofbits", "product": "fickling", "versions": [{"status": "affected", "version": "0", "versionType": "custom", "lessThanOrEqual": "0.1.11"}, {"status": "unaffected", "version": "0.1.12", "versionType": "custom"}], "packageName": "fickling", "collectionURL": "https://pypi.org/project/fickling/", "defaultStatus": "unaffected"}]}], "published": "2026-07-04T14:16:29.063", "references": [{"url": "https://github.com/trailofbits/fickling/commit/41ce7cb01edd97072994039574a2301ebb3f463d", "tags": ["Patch"], "source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082"}, {"url": "https://github.com/trailofbits/fickling/pull/278", "tags": ["Issue Tracking", "Patch"], "source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082"}, {"url": "https://github.com/trailofbits/fickling/releases/tag/v0.1.12", "tags": ["Release Notes"], "source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082"}, {"url": "https://github.com/trailofbits/fickling/security/advisories/GHSA-cffv-grgg-g429", "tags": ["Exploit", "Third Party Advisory"], "source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082"}, {"url": "https://github.com/trailofbits/fickling/security/advisories/GHSA-cffv-grgg-g429", "tags": ["Exploit", "Third Party Advisory"], "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0"}], "vulnStatus": "Analyzed", "weaknesses": [{"type": "Secondary", "source": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082", "description": [{"lang": "en", "value": "CWE-693"}]}], "descriptions": [{"lang": "en", "value": "In Trail of Bits fickling versions up to and including 0.1.11, the UnsafeImportsML analysis pass unconditionally calls AnalysisContext.shorten_code(node) on every import node it inspects, regardless of whether the import is flagged as unsafe. This call registers the shortened code representation in the shared AnalysisContext.reported_shortened_code set. When the MLAllowlist analysis pass subsequently runs, it calls the same shorten_code() method, receives already_reported=True for every import, and executes a continue statement that skips its allowlist check entirely. This renders MLAllowlist dead code for all imports \u2014 it never evaluates whether an import is in the ML allowlist or not. The MLAllowlist pass was designed to catch imports of modules outside the known-safe ML ecosystem (torch, numpy, transformers, etc.) that slip past the UnsafeImports denylist. With MLAllowlist inoperative, any standard library module not in the UNSAFE_IMPORTS denylist can be invoked via pickle deserialization while fickling's check_safety() returns LIKELY_SAFE. The fickling.load() API chains check_safety() into pickle.loads() as an explicit security gate, meaning a LIKELY_SAFE verdict causes the payload to be deserialized and executed. The root cause is shared mutable state between independently-correct analysis passes \u2014 UnsafeImportsML works as designed in isolation, MLAllowlist works as designed in isolation, but the shared reported_shortened_code set causes UnsafeImportsML to poison MLAllowlist's deduplication logic."}, {"lang": "es", "value": "En las versiones de fickling de Trail of Bits hasta la 0.1.11 inclusive, la pasada de an\u00e1lisis UnsafeImportsML llama incondicionalmente a AnalysisContext.shorten_code(node) en cada nodo de importaci\u00f3n que inspecciona, independientemente de si la importaci\u00f3n est\u00e1 marcada como insegura. Esta llamada registra la representaci\u00f3n de c\u00f3digo acortada en el conjunto compartido AnalysisContext.reported_shortened_code. Cuando la pasada de an\u00e1lisis MLAllowlist se ejecuta posteriormente, llama al mismo m\u00e9todo shorten_code(), recibe already_reported=True para cada importaci\u00f3n y ejecuta una sentencia continue que omite completamente su verificaci\u00f3n de lista de permitidos. Esto convierte a MLAllowlist en c\u00f3digo muerto para todas las importaciones - nunca eval\u00faa si una importaci\u00f3n est\u00e1 en la lista de permitidos de ML o no. La pasada MLAllowlist fue dise\u00f1ada para detectar importaciones de m\u00f3dulos fuera del ecosistema de ML conocido como seguro (torch, numpy, transformers, etc.) que escapan a la lista de denegados de UnsafeImports. Con MLAllowlist inoperativa, cualquier m\u00f3dulo de la biblioteca est\u00e1ndar que no est\u00e9 en la lista de denegados UNSAFE_IMPORTS puede ser invocado a trav\u00e9s de la deserializaci\u00f3n de pickle mientras que check_safety() de fickling devuelve LIKELY_SAFE. La API fickling.load() encadena check_safety() en pickle.loads() como una puerta de seguridad expl\u00edcita, lo que significa que un veredicto LIKELY_SAFE hace que la carga \u00fatil sea deserializada y ejecutada. La causa ra\u00edz es el estado mutable compartido entre pasadas de an\u00e1lisis independientemente correctas - UnsafeImportsML funciona seg\u00fan lo dise\u00f1ado de forma aislada, MLAllowlist funciona seg\u00fan lo dise\u00f1ado de forma aislada, pero el conjunto compartido reported_shortened_code hace que UnsafeImportsML envenene la l\u00f3gica de deduplicaci\u00f3n de MLAllowlist."}], "lastModified": "2026-10-25T15:10:00.147", "configurations": [{"nodes": [{"negate": false, "cpeMatch": [{"criteria": "cpe:2.3:a:trailofbits:fickling:*:*:*:*:*:python:*:*", "vulnerable": true, "matchCriteriaId": "A3AF23ED-3F44-48AF-A1AC-CFCDC22A55D3", "versionEndIncluding": "0.1.11"}], "operator": "OR"}]}], "sourceIdentifier": "aa17e1a1-c329-4d6e-a1ed-8d0188aea082"}