CVE-2026-73067
Description
A flaw was found in Tesseract, an open-source Optical Character Recognition (OCR) engine. A remote attacker could exploit this vulnerability by providing a specially crafted .traineddata model. When this malicious model is loaded, it can cause the software to read beyond its allocated memory, leading to a process crash. This results in a denial of service, making the Tesseract application unavailable.
Statement
A flaw was found in Tesseract, an open-source Optical Character Recognition (OCR) engine. A heap out-of-bounds read in the DAWG dictionary loader can be triggered by a specially crafted .traineddata model file, causing the Tesseract process to crash. This issue requires a user to load an untrusted model file, limiting the attack surface.
Mitigation
Only use .traineddata model files obtained from trusted sources. Do not load model files from untrusted or unknown origins.
Frequently Asked Questions
Not sure what something means? Check out our Security Glossary.
Want to get errata notifications? Sign up here.
