pandas‑ai version 3.0.0 contains a flaw in the _execute_sql_query component that permits arbitrary SQL injection. By supplying a crafted query string, an attacker can modify the database command executed, which can lead to unauthorized reading, updating, or deleting of data. The vulnerability is a classic example of CWE-89, where unsanitized input is used to compose database queries.

The affected module is the pandasai library developed by Gabriele Eventuri, specifically version 3.0.0 for Python. Applications that import pandasai and expose the _execute_sql_query method to user‑provided data are at risk. Any system that relies on this library to run database queries could be compromised if it does not filter or treat input properly.

The CVSS score of 7.3 indicates high severity, yet the EPSS score of less than 1% suggests that exploitation is currently unlikely. The vulnerability is not listed in the CISA KEV catalog. The likely attack vector is an attacker influencing the query string passed to _execute_sql_query, which is inferred from the description and is typical for web or service applications that accept user‑sourced SQL statements.