A predictable resource name flaw in the BigQuery Import Staging component of Google Cloud Vertex AI Search for Commerce allows an attacker who knows the victim’s project number to construct the staging bucket name and obtain read or write access to staged data and error logs. The flaw results in the disclosure of sensitive data, the potential tampering or deletion of process logs, and broader compromise of data confidentiality and integrity, as reflected by the CVSS score of 9.3.

Vertex AI Search for Commerce deployments on Google Cloud Platform running any version before the 2026‑04‑27 release are impacted. The issue surfaces when the product creates stable, predictable bucket names for staging, making the buckets discoverable by computing the name from the public project number.

The vulnerability carries a severe risk rating of 9.3 on the CVSS scale and is considered a high‑impact resource name exposure. It is not listed in the CISA KEV catalog and the EPSS score is not available. Exploitation requires only knowledge of the target project number and the ability to perform GCS operations on the target bucket; no additional authentication is needed beyond bucket access controls. Because the vendor has released a patch that eliminates the predictable naming pattern, no further action is required beyond ensuring the patch is in place.