Broadcom has disclosed a sweeping security advisory covering 91 flaws in the Spring Framework and associated open-source projects, impacting more than 209,000 software components. These vulnerabilities were identified by Sonatype and revealed in a Broadcom advisory published August 20, 2026. The scope of the issues spans widely deployed Spring modules such as Spring Security, Cloud Config, AI, Data REST, Integration, Reactor Core, Reactor Netty, AMQP, and Batch. Because multiple version lines are affected, organizations may need to update several branches rather than a single release.
Major Flaws & Risk Vectors
Among the vulnerabilities, insecure deserialization is a standout concern. The critical CVE-2026-59285 flaw in Spring for GraphQL risks remote code execution when Jackson 2.x is used for JSON deserialization, paginated GraphQL fields are exposed, and unsafe classes are deserialized. The severity score for this issue is rated 9.2.
Another serious issue, CVE-2026-59318, arises in the Spring AI “tool-calling” features. It could enable a prompt injection attack, letting an attacker invoke tools not intended for certain requests. In systems where permissions are not tightly enforced at the system layer, that could lead to privilege escalation. Other disclosed problems include server-side request forgery, path traversal, denial-of-service conditions, information leakage, and authorization weaknesses.
Downstream Impact & Mitigation Challenges
The reach of these Spring vulnerabilities is amplified by how open-source dependencies are incorporated. Many affected packages don’t directly import Spring but include it transitively via upstream frameworks or embedded libraries. Even when patches are issued, enterprises must execute a chain of processes—upgrading library versions, ensuring maintainers adopt fixed releases, rebuilding software, and deploying those new builds—for security to improve.
Sonatype’s research points to a sharp uptick in vulnerability discovery, partly driven by AI tools analyzing extensive codebases. The newly affected component versions grew roughly 46 times over the pre-AI discovery rate. In parallel, the number of critical or high‐severity vulnerabilities per enterprise app rose by roughly 4.3× over the last four years. Also, reports of Spring security advisories leapt by more than 1,700% between March and April 2026.
What Enterprises Should Do Now
Organizations using Spring, especially internet-exposed services, GraphQL, or AI components—or handling untrusted input—should immediately map their dependency footprint, including transitive dependencies. They should assess exposure via reachable attack paths more than simply tallying CVE counts. Patch to fixed versions as listed in the advisories and run thorough tests to ensure stability.
Scanning software bills of materials (SBOMs), validating both direct and transitive dependencies, and conducting threat modeling focusing on components with public exposure or AI integrations are also essential. Surfacing where Spring code underlies upstream frameworks is crucial to understanding full risk.
This wave of Spring flaws underscores both the power and complexity of open-source ecosystems in the age of AI-assisted research. The disclosure not only reflects a maturing of research tools detecting vulnerabilities faster, but also reveals an urgent need for mature dependency intelligence and faster remediation pipelines. Teams that can’t move as swiftly risk having flaws linger in production—especially in components long assumed safe because they were inherited transitively. What comes next won’t just be new fixes—it’ll be who moves first, who automates best, and who shapes that velocity into resilient architectures.
