What is the CVSS score of CVE-2026-44018?

CVE-2026-44018 has a CVSS 3.1 base score of 7.1 (High). CVSS vector: CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:H. EPSS exploitation probability: 0.0%.

Which versions of Docling are affected by CVE-2026-44018?

Docling, a Python document conversion library in versions 2.45.0-2.91.0, contains vulnerabilities that enable reading of sensitive local files or denial of service when processing malicious…. A patch is available.

How to fix or mitigate CVE-2026-44018?

Within 24 hours: Identify all production systems running Docling 2.45.0-2.91.0 and confirm whether they process untrusted documents. Within 7 days: Deploy vendor patch to all affected systems. Within 30 days: Complete global patch rollout, validate patching completeness, and search application logs for exploitation attempts.

Docling CVE-2026-44018

EUVD-2026-39790 HIGH

Improper Handling of Highly Compressed Data (Data Amplification) (CWE-409)

2026-06-03 https://github.com/docling-project/docling

GHSA-r3xg-rg9j-67fv

Denial Of Service XXE Red Hat

7.1

CVSS 3.1 · NVD

Severity by source

NVD PRIMARY

7.1 HIGH

AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:H

vuln.today AI

7.1 HIGH

Malicious input must be processed locally (AV:L, UI:R) with no privileges (PR:N); XXE yields high file-read confidentiality (C:H) and bombs cause DoS (A:H), with no integrity impact (I:N).

3.1 AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:H

4.0 AV:L/AC:L/AT:N/PR:N/UI:P/VC:H/VI:N/VA:H/SC:N/SI:N/SA:N

Red Hat

5.5 MEDIUM

qualitative

Primary rating from NVD.

CVSS VectorNVD

Attack Vector

Local

Attack Complexity

Low

Privileges Required

None

User Interaction

Required

Scope

Unchanged

Confidentiality

High

Integrity

None

Availability

High

Lifecycle Timeline

Analysis Updated

Jun 27, 2026 - 20:43 vuln.today

v3 (cvss_changed)

Analysis Updated

Jun 27, 2026 - 20:43 vuln.today

v2 (cvss_changed)

Re-analysis Queued

Jun 27, 2026 - 20:37 vuln.today

cvss_changed

Severity Changed

Jun 27, 2026 - 20:37 NVD

MEDIUM HIGH

CVSS changed

Jun 27, 2026 - 20:37 NVD

5.5 (MEDIUM) 7.1 (HIGH)

Source Code Evidence Fetched

Jun 03, 2026 - 21:51 vuln.today

Analysis Generated

Jun 03, 2026 - 21:51 vuln.today

DescriptionNVD

Impact

The METS-GBS backend's XML parsing and the input document format detection lacked security controls, enabling:

XML External Entity (XXE) attacks to read local files or cause denial of service
Decompression bombs (zip bombs) to exhaust memory and disk space
Unbounded archive extraction consuming system resources

An attacker could craft malicious METS-GBS archives that, when processed, could read sensitive files, exhaust system resources, or cause application crashes.

Patches

Fixed in version 2.91.0. The fix implements:

Secure XML parsing with resolve_entities=False, load_dtd=False, and no_network=True
Configurable limits: 300 MB total extraction size, 10 MB per file, 1000 member count
Cumulative size tracking across all extractions
Early termination when limits are exceeded
Secure format detection of METS-GBS tar archives with _detect_mets_gbs() method: maximum file size (10 MB per file), maximum member count (1000 members), and exception handling to gracefully fail when limits are exceeded

Workarounds

Avoid processing METS-GBS archives from untrusted sources. If necessary, pre-validate archives in an isolated environment with resource limits.

References

Fix release: v2.91.0

AnalysisAI

Local file disclosure and denial of service in the Docling document-conversion library (Python pip package, versions 2.45.0 up to 2.91.0) stem from its METS-GBS backend parsing XML and extracting archives without security controls. A maliciously crafted METS-GBS archive processed by Docling can trigger XML External Entity (XXE) resolution to read sensitive local files, or use decompression/zip bombs and unbounded tar extraction to exhaust memory and disk. EPSS is very low (0.01%, 3rd percentile) and there is no public exploit identified at time of analysis; exploitation requires a victim application to process attacker-supplied input.

Technical ContextAI

Docling is a popular open-source Python library (distributed as pkg:pip/docling) used to parse and convert documents - increasingly in RAG and LLM ingestion pipelines. The flaw lives in the METS-GBS backend, which handles METS-GBS tar archives (a METS-based packaging format for digitized books/Google Books-style content). Two root causes combine: an unsafe XML parser that resolved external entities and DTDs (classic XXE) and an archive extractor with no size/member limits. The assigned CWE-409 (Improper Handling of Highly Compressed Data / decompression bomb) captures the resource-exhaustion half; the advisory and tags also explicitly cite XXE, which is a CWE-611 class issue. The fix in 2.91.0 hardens XML parsing with resolve_entities=False, load_dtd=False, and no_network=True, and adds a _detect_mets_gbs() detector plus extraction limits (300 MB total, 10 MB per file, 1000 members) with cumulative tracking and early termination.

RemediationAI

Vendor-released patch: 2.91.0 - upgrade docling to 2.91.0 or later (pip install --upgrade 'docling>=2.91.0'), which enforces secure XML parsing (resolve_entities=False, load_dtd=False, no_network=True) and extraction limits (300 MB total, 10 MB per file, 1000 members). If you cannot upgrade immediately, the vendor workaround is to avoid processing METS-GBS archives from untrusted sources; where such processing is required, pre-validate archives in an isolated sandbox with strict CPU, memory, and disk quotas (the trade-off is added pipeline latency and operational overhead) and disable or gate the METS-GBS ingestion path so only trusted, size-bounded inputs reach the backend. See the advisory https://github.com/docling-project/docling/security/advisories/GHSA-r3xg-rg9j-67fv and release notes https://github.com/docling-project/docling/releases/tag/v2.91.0 (fix PR #3336, commit c1dbac2).