CVE-2026-44017
HighCVSS 7.5Exploitation Probability (EPSS)
Low risk49th percentile - higher than 49% of all known CVEs
Summary
A vulnerability in Docling before version 2.91.0 allows Zip Slip attacks during EasyOCR model downloads. An attacker can overwrite arbitrary files, leading to remote code execution, persistent backdoors, or data corruption.
Risk Assessment
The risk includes full server compromise via remote code execution, persistent backdoors, and data integrity loss, potentially resulting in complete system breach.
Recommendation
Upgrade Docling to version 2.91.0 or later immediately. Additionally, secure model download sources against supply chain attacks, DNS spoofing, and MITM attacks.
Other vulnerabilities in Docling
See all- CVE-2026-47214High
In Docling prior to version 2.94.0, unsafe URI and path handling was found in the HTML backend. This vulnerability could allow an attacker to perform unauthorized operations on files or network resources.
- CVE-2026-44020High
A vulnerability in Docling library versions 2.13.0 through 2.74.0 allows XML External Entity (XXE) attacks via parsing malicious USPTO patent XML files. An attacker can read arbitrary files from the server, perform SSRF attacks, or cause denial of service through entity expansion.
- CVE-2026-44016High
In Docling versions 2.82.0 through 2.90.0, if the HTML backend was explicitly configured for rendering (disabled by default), the Playwright-based rendering feature could allow JavaScript execution and unrestricted network access when processing untrusted HTML documents. An attacker could craft malicious HTML that executes arbitrary JavaScript in the rendering context or makes unauthorized network requests to internal services.
Original NVD description (English source)
Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. Prior to 2.91.0, the EasyOCR model download functionality extracted ZIP archives without validating member paths, enabling Zip Slip attacks. If an attacker could compromise the model download source (via supply chain attack, DNS spoofing, or MITM), they could write arbitrary files to any location writable by the process, potentially achieving remote code execution by overwriting Python files or system binaries, persistent backdoors by modifying startup scripts or SSH keys, and data corruption or system compromise. This vulnerability is fixed in 2.91.0.