91 New Spring CVEs Just Hit 209,000+ Software Components

On August 20, 2026, Broadcom released one of the largest coordinated vulnerability disclosures the Java ecosystem has seen in years. Sonatype tracked 91 CVEs across the Spring Framework and its satellite projects, impacting an estimated 209,569 software components — and that number was still climbing at the time of publication.

The affected projects span Spring Security, Spring Cloud Config, Spring AI, Spring Data REST, Spring Integration, Reactor Core, Reactor Netty, Spring AMQP, and Spring Batch. The flaw categories include insecure deserialization, untrusted code execution, information disclosure, server-side request forgery, path traversal, denial-of-service, and broken authorization logic.

Two CVEs stand out from the rest. CVE-2026–59285 is a 9.2 CRITICAL unsafe deserialization vulnerability in Spring for GraphQL. When an application uses Jackson 2.x for JSON deserialization, exposes paginated GraphQL fields, and has certain classes reachable during deserialization, an attacker could potentially achieve remote code execution.

CVE-2026–59318 affects Spring AI's tool-calling functionality. Under specific conditions, a prompt-injection attack can trick the system into invoking a tool that was never intended for that request, opening a path toward privilege escalation. This is one of the first mainstream previews of a new vulnerability class sitting at the intersection of traditional software security and AI agent behavior.

Here's what makes this disclosure especially difficult to triage: most organizations don't import Spring directly in every service. It arrives buried two or three layers deep through internal SDKs, logging wrappers, and other transitive dependencies. A patched upstream release doesn't automatically protect you — maintainers have to adopt it, teams have to rebuild, and pipelines have to redeploy.

This disclosure also reflects a much larger shift in how vulnerabilities are being found. Sonatype's research noted that newly affected component versions are appearing at roughly 46 times the pre-AI rate, while Broadcom separately reported a 1,700%+ increase in monthly Spring security advisories between March and April 2026 alone. AI-assisted vulnerability research is accelerating disclosure at a pace no manual review process can match.

For security teams, the practical priority list looks like this: patch by exposure, not by CVE count. Lock down GraphQL endpoints running Jackson 2.x deserialization first. Enforce authorization at the system layer for any Spring AI implementation, not just the tool-selection layer. Rebuild and redeploy affected services rather than just updating a version tag. And validate every upgrade through your test suite, since Spring's advisory portal lists fixes across multiple supported version branches.

I put together a full technical breakdown covering every affected project, both critical CVEs in detail, SBOM detection commands using Syft and Grype, dependency-tree commands for Maven and Gradle, and a prioritized remediation checklist built specifically for SOC and DevSecOps teams.

Read the complete breakdown here: https://www.xpert4cyber.com/2026/08/91-spring-vulnerabilities-cve-2026.html

If your team runs anything on Spring Boot, Spring Security, or Spring Cloud, this is worth fifteen minutes of your time today — not next sprint.