CVE-2026-105748

MediumCVSS 4.3/10Last modified Oct 7, 2026

Description

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.16.0 until 2.131.0, the InputFormat.JSON_DOCLING backend in docling/backend/json/docling_json_backend.py validates serialized DoclingDocument input without rejecting picture image references that contain local paths or file URIs. When the document is enriched or exported with ImageRefMode.EMBEDDED, the DoclingDocument._with_embedded_pictures and ImageRef.pil_image methods can open those references and place readable image bytes in Markdown or HTML output. Disclosure is limited to files Pillow can decode as images, while differing decode behavior can also reveal whether a path exists. Direct untrusted loading through docling-core is outside this Docling fix. This issue is fixed in 2.131.0.

Metrics

CVSS 3.1 4.3/10CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:N/A:N

Weakness enumeration

CWE-73CWE-200

Affected software

No affected versions have been published.

Timeline

Published
Oct 5, 2026
Last modified
Oct 7, 2026
NVD status
Analyzed

Related CVEs from 2026

CVE-2026-1057496.5Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.0.0 unCVE-2026-105758.8IBM MQ could allow an authenticated attacker to cause a denial of service or potentially escalate privileges due to a heap buffer overflow wCVE-2026-1057507.5Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.82.0 uCVE-2026-1057517.5Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.107.0 CVE-2026-1057523.1vLLM is an inference and serving engine for large language models. Prior to 0.30.0, Harmony tool continuations submitted through "POST /v1/rCVE-2026-1057536.5vLLM is an inference and serving engine for large language models. Prior to 0.28.0, the default mirrored multimodal LRU cache can commit a mCVE-2026-1057546.5vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the /inference/v1/generate endpoint in the disaggregatedCVE-2026-1057554.2vLLM is an inference and serving engine for large language models. Prior to 0.30.0, flash late-interaction scoring at the /score and /rerankCVE-2026-1057566.5vLLM is an inference and serving engine for large language models. Prior to 0.30.0, OpenAI-compatible request models accept a non-empty cachCVE-2026-1057576.5vLLM is an inference and serving engine for large language models. Prior to 0.30.0, structured-output request failures can escape request-scCVE-2026-1057585.3vLLM is an inference and serving engine for large language models. From 0.24.0 until 0.30.0, the Qwen2VLVideoBackend and Qwen3VLVideoBackendCVE-2026-1057597.5vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the Rust frontend's track_http_metrics middleware record
CVE-2026-105748: Medium severity, CVSS 4.3 | Vulnix