AWS Fixes AutoGluon Dataset Preparation Path Traversal

P2 — VALIDATE AND UPDATEModerate · 6.8 (CVSS v3.1)EXPLOITATION: NOT STATED

Quick Answer

Upgrade the optional AutoGluon dataset preparation utility to 1.6.0 or later and review prior untrusted archive extraction where warranted.

Confirm Scope And Apply The Supported Remediation

What to do now: Find benchmark-data preparation jobs, notebooks, runners and forks that invoke prepare_detection_dataset, including the account privileges used. Confirm use of the optional utility and its download path. Do not infer exposure merely from AutoGluon training or inference API use. Upgrade autogluon.multimodal and maintained forks. If blocked, stop using the utility and follow AWS guidance for HTTPS downloads and path-safe extraction. Confirm installed versions and test a trusted benchmark dataset. Retain dataset provenance, extraction paths and relevant job logs.

Open the authoritative advisory

Last verified: 2026-11-19 UTC. Recheck the authoritative advisory and supported distribution channel before changing production.

Scope And Authority

Product scopeAutoGluon MultiModal optional object-detection dataset preparation utility
Advisory2026-188-AWS, GHSA-cx2g-xwj2-2vxm
CVECVE-2026-130638
Authoritative release date2026-11-18T18:00:00Z — AWS bulletin states 10:00 AM PST.
Authority revision dateNo separate material revision is stated for this release.
Affected versionsautogluon.multimodal before 1.6.0 when the optional prepare_detection_dataset utility downloads benchmark archives over plaintext HTTP.
Fixed versionautogluon.multimodal 1.6.0 or later
CVSS base score6.8 (CVSS v3.1)
CVSS severityModerate
Exploitation statusThe cited authority does not confirm exploitation of these vulnerabilities in the wild.; no exploitation claim is inferred.

What Changed

An on-path attacker could substitute a dataset archive that writes outside the extraction directory with the utility user’s privileges. AWS explicitly excludes MultiModalPredictor training and inference APIs from this utility’s scope.

What To Validate Now

  1. Inventory. Find benchmark-data preparation jobs, notebooks, runners and forks that invoke prepare_detection_dataset, including the account privileges used.
  2. Establish applicability. Confirm use of the optional utility and its download path. Do not infer exposure merely from AutoGluon training or inference API use.
  3. Remediate. Upgrade autogluon.multimodal and maintained forks. If blocked, stop using the utility and follow AWS guidance for HTTPS downloads and path-safe extraction.
  4. Validate. Confirm installed versions and test a trusted benchmark dataset. Retain dataset provenance, extraction paths and relevant job logs.
  5. Retain evidence. Preserve asset and owner identifiers, the applicability decision, before-and-after versions, change approval, installation output, validation results, and any exception or rollback record.

Operational Cautions

Separate the utility’s filesystem privileges from model-serving access. A successful upgrade does not establish that previously extracted files are trustworthy.

Evidence To Retain

  • Exact product, release stream, package or application version, enabled feature, environment, and accountable owner.
  • UTC timestamps, authority revision, approved change record, installer or package-manager output, and resulting version.
  • Relevant logs and monitoring evidence, test results, exceptions, compensating controls, and rollback decisions.

Related TechGeeks Resources

The AWS-maintained AutoGluon repository advisory supplies the CVSS v3.1 base score of 6.8 and Moderate severity.

Authoritative References

Correction policy: If the authority changes affected versions, fixed versions, severity, exploitation information, mitigations, or required action, TechGeeks will update this notice and its verification date.