AWS SDK for pandas Iceberg write SQL identifier flaw lets users read other tables

A Python data tool for AWS can let people peek into other users' tables.

Summary

A flaw in how the Apache Iceberg write component builds SQL identifiers lets an authenticated user read tables that a more privileged principal can access. Versions 3.5.2 through 3.17.0 of AWS SDK for pandas (awswrangler) are affected.

Affected when

An authenticated user with permission to modify AWS Glue Data Catalog metadata
A more privileged principal calling to_iceberg() to write to a table

What to do

Update to 3.17.1 or later. Until then, avoid calling to_iceberg() with a principal that has more privileges than those who can modify the table's metadata. Also restrict glue:UpdateTable and related permissions on shared databases.

Products and platforms

AWS SDK for pandas

CVE IDs

CVE-2026-119301 (opens in a new tab)

Sources

Advisory: github.com (opens in a new tab) · aws.amazon.com (opens in a new tab)
Write-up: github.com (opens in a new tab)

Times

First published 2026-11-02 16:34 UTC

First seen 2026-11-09 14:10 UTC