AmazonAnalysisAWS SDK for pandas patches high-severity Iceberg SQL identifier injection
What happened.Amazon Web Services issued security bulletin 2026-166-AWS detailing an improper quoting vulnerability in the AWS SDK for pandas Apache Iceberg write pathway.
Why it matters.Cloud data pipelines run ETL transformations using high-privilege service roles to process analytics. When metadata attributes inject raw SQL identifiers into Iceberg transactions, lower-privileged users escalate access into protected tables. Data platform engineers must upgrade awswrangler packages across scheduled pipelines. Security teams should enforce strict least-privilege IAM policies on AWS Glue Data Catalog update operations.
The technical specifications:
- Vulnerability mechanics: Tracked as CVE-2026-119301 with a CVSS score of 7.9, the flaw stems from unneutralized quoting syntax during SQL identifier generation in to_iceberg operations.
- Privilege escalation: Authenticated users with AWS Glue Data Catalog edit permissions can craft malicious column names that alter table schemas during automated pipelines.
- Unauthorized access: Exploiting the defect enables unauthorized principals to read restricted tables or overwrite data across shared warehouse partitions.
- Affected versions: The vulnerability impacts awswrangler versions 3.5.2 through 3.17.0 across Python data engineering environments.
- Remediation release: AWS released awswrangler version 3.17.1, advising teams to upgrade libraries and restrict AWS Glue catalog update permissions.