CVE-2026-42440 Apache Software Foundation CVE debrief

Who should care

Organizations using Apache OpenNLP versions before 1.9.5, 2.5.9, or 3.0.0-M3 should prioritize patching this vulnerability. This includes developers and administrators who work with machine learning models, especially those loaded from untrusted or semi-trusted sources. Patching is crucial to prevent potential denial-of-service attacks that could impact service availability.

Technical summary

The vulnerability lies in the AbstractModelReader methods getOutcomes(), getOutcomePatterns(), and getPredicates(), which read a 32-bit signed integer count field from a binary model stream and pass it directly to array allocations without validation. This allows an attacker to control the count fully when the model file comes from an untrusted source. A crafted .bin model file with a count field set to Integer.MAX_VALUE or a large enough value to exhaust the available heap triggers an OutOfMemoryError at array allocation. The error occurs early in deserialization, making it easy for an attacker to weaponize a payload with a small file. Any code path deserializing a .bin model is affected, including direct use of GenericModelReader and higher-level components that delegate to it during model load.

Defensive priority

High priority should be given to patching this vulnerability, especially in environments where model files are loaded from untrusted or semi-trusted sources. Immediate action is recommended for versions before 1.9.5, 2.5.9, and 3.0.0-M3.

Recommended defensive actions

• Upgrade to version 2.5.9 for 2.x users.
• Upgrade to version 3.0.0-M3 for 3.x users.
• Treat all .bin model files as untrusted input unless their provenance is verified.
• Avoid loading models supplied by end users or fetched from third-party repositories without integrity checks.
• Consider setting the OPENNLP_MAX_ENTRIES system property to a desired positive integer for deployments that need to load models with more entries than the default.

Evidence notes

The CVE-2026-42440 vulnerability was publicly disclosed on May 4, 2026, and modified on June 29, 2026. The vulnerability affects Apache OpenNLP versions before 1.9.5, 2.5.9, and 3.0.0-M3. The CVSS score is 7.5 with a HIGH severity. The vulnerability allows for a denial-of-service attack via OOM.

Official resources