The vulnerability in the Athena module of AWS SDK for pandas (awswrangler) allows an unauthenticated or remote actor to read query results and manipulate the data returned to the caller. The root cause is that the default query result location uses a pre‑created S3 bucket whose ownership is not verified, enabling a malicious actor with read/write permissions on that bucket to intercept or alter query outputs. This flaw aligns with CWE‑283, where an attacker abuses account or authentication token privileges to access resources beyond their intended scope.

The issue affects the AWS SDK for pandas package from AWS (awswrangler) itself. All installed versions prior to 3.16.1 are impacted. Users should verify that the default Athena result bucket in each account and region belongs to their own account or update the library to 3.16.1 or later, during which the verification logic was added.

The CVSS score of 5.9 indicates moderate severity. No EPSS score is currently available, and the vulnerability is not listed in the CISA KEV catalog, suggesting limited public exploitation data. The likely attack vector is remote and requires the attacker to have access to the S3 bucket used as the Athena result location, which may happen if the bucket is shared or misowned. Proper ownership verification or explicit result configuration reduces the risk, but until the library is updated or the bucket ownership corrected, the risk remains present.