Unbounded Tika text extraction can exhaust JVM memory

HIGH | NOVEMBER 19, 2026 | CVE-2026-59338

Description

TikaDocumentReader in Spring AI uses Apache Tika to extract text from documents. Its default configuration disables Tika's built-in content-size limit, so the extracted text of a parsed document is buffered in memory with no absolute bound. Apache Tika still aborts a parse when the extracted text is disproportionate to the size of the input, which rejects highly compressed inputs such as archive bombs, but that guard is relative to the input and places no ceiling on the total amount of text extracted. An application that uses TikaDocumentReader to process documents from an untrusted source (e.g., user-uploaded files or attacker-supplied URLs) may therefore be induced to allocate an excessive amount of memory while parsing a sufficiently large document, potentially exhausting available JVM heap and causing a denial of service.

Affected Spring Products and Versions

Spring AI:

  • 2.0.0 - 2.0.2
  • 1.1.0 - 1.1.9
  • 1.0.0 - 1.0.10

Mitigation

Users of affected versions should upgrade to the corresponding fixed version.
Fix versionAvailability
2.0.3OSS
2.0.2.1Enterprise Support Only
1.1.10Enterprise Support Only
1.0.11Enterprise Support Only

No further mitigation steps are necessary.

Credit

The issue was reported responsibly by aleister1102 (https://github.com/aleister1102) and independently reported by Raphael Zanarelli - https://github.com/zanarellidev

History

  • 2026-11-19: Initial vulnerability report published.

Get ahead

VMware offers training and certification to turbo-charge your progress.

Learn more

Get support

TrueSource Spring Enterprise offers support and binaries for OpenJDK™, Spring, and Apache Tomcat® in one simple subscription.

Learn more

Upcoming events

Check out all the upcoming events in the Spring community.

View all