Skip to content
Vulnerabilities

Amazon Neptune graph import flaw risks credential theft

A critical deserialization bug in Amazon Neptune allows authenticated users to execute arbitrary code and steal internal service credentials.

Illustration: A sleek graph database node sits isolated on a clean white surface.
AI-generated illustration: AppSec Scoop

Key points

  • CVE-2026-110812 affects Amazon Neptune engine versions from 1.2.0.0 up to 1.5.0.0.
  • The flaw resides in the Gremlin graph data import component within the query engine.
  • Patches are available for engine lines 1.4.7, 1.4.8 and 1.5.0.

Amazon Neptune engine versions prior to 1.5.0.0 contain a critical deserialization flaw in the Gremlin graph import component. An authenticated actor with standard database permissions can exploit this issue to execute arbitrary code and access internal service credentials.

What happened

AWS published security bulletin 2026-136-AWS on 15 October 2026 to address CVE-2026-110812. The National Vulnerability Database rates the issue as critical with a CVSS score of 9.9. The weakness stems from unsafe processing of untrusted input within the Apache TinkerPop Gremlin functionality used by the service.

The flaw allows an authenticated principal with Neptune data-plane permissions to submit requests that cause the query engine to process input unsafely. This behaviour could lead to unintended activity within the query engine environment. Attackers could potentially execute arbitrary code and gain access to internal service credentials.

AWS states that the issue has been addressed in engine versions 1.4.7.1, 1.4.8.1 and 1.5.0.0. Customers who manage engine versions manually should upgrade their clusters to these patched versions. Those enrolled in auto-update will receive the fix during their next scheduled maintenance window.

Why it matters

The vulnerability requires an authenticated user with data-plane access, meaning attackers must first obtain valid credentials. However, the potential for arbitrary code execution and credential theft remains severe. This exposure affects any organisation using Amazon Neptune for graph database operations.

Security teams must verify which engine versions are running in their environments. Systems running versions from 1.2.0.0 up to 1.5.0.0 are impacted. The ability to access internal service credentials could allow lateral movement within the cloud environment.

What happens next

  • Identify all Amazon Neptune clusters running vulnerable engine versions.
  • Schedule upgrades to patched versions 1.4.7.1, 1.4.8.1 or 1.5.0.0.
  • Restrict Neptune data-plane permissions to only trusted principals.
  • Monitor for unusual activity within the query engine environment.

Patch priority

Teams should prioritise updating clusters that expose the Gremlin import path to external or semi-trusted users. Systems with broad data-plane permissions pose the highest risk and require immediate attention. Before applying updates, validate that backups are current and test the upgrade in a non-production environment. This approach minimises the risk of service disruption while ensuring the patch applies cleanly.

What to do and how to stay safe: Amazon Neptune

  • Audit IAM policies to ensure only trusted principals have Neptune data-plane permissions.
  • Review application logs for unusual query patterns or unexpected data imports.
  • Verify that auto-minor-version upgrades are enabled for future automatic fixes.
  • Isolate affected clusters from untrusted networks until patches are applied.

Step-by-step guide: Patch Management: Definition, Purpose and Operational Reality

General security guidance from the AppSec Scoop newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Which Amazon Neptune versions are affected by CVE-2026-110812?

Engine versions from 1.2.0.0 up to 1.5.0.0 are affected by this deserialization flaw.

Is there a patch available for this vulnerability?

Yes, AWS has released patches in engine versions 1.4.7.1, 1.4.8.1 and 1.5.0.0.

What permissions are required to exploit this flaw?

An attacker needs authenticated access with Neptune data-plane or query permissions.

Sources

  1. CVE Program
  2. AWS Security Bulletins
Amazon NeptuneCVE-2026-110812AWSGremlindeserialization

Related stories