CVE-2025-12397

Root

Cause

A SQL injection vulnerability exists in Looker Studio when BigQuery is used as the data source [2]. The root cause is improper sanitization of user-controlled input within the batchedDataV2 HTTP request. When a user accesses a report, dynamically generated column aliases are directly injected into the underlying BigQuery query. An attacker can manipulate these aliases to inject arbitrary SQL commands, bypassing limitations such as the removal of dots and spaces [2].

Exploitation

A Looker Studio user with only report view-level access can exploit this vulnerability [1][2]. The attack requires the victim to visit the attacker's malicious report. The attacker then proxies the victim's requests, intercepts the batchedDataV2 request, and injects malicious SQL into the JSON values of that request [2]. The crafted payload is executed when the legitimate user's session processes the report.

Impact

Successful exploitation executes arbitrary SQL queries under the report owner's identity and permissions [1][2]. This could lead to exfiltration, modification, or deletion of any data the owner has access to within BigQuery, depending on the owner's database privileges [2].

Mitigation

This vulnerability was patched by Google on 21 July 2025 [1]. No customer action is required, as the fix was applied server-side [1].