An insecure implementation of the is_sql_query_safe function in PandasAI’s sql_sanitizer module permits malicious manipulation of the query string to cause the application to resolve file paths outside the intended directory, enabling path traversal. This flaw is classified as CWE‑22 and can lead to unauthorized read or write access to files on the host system, potentially compromising confidentiality, integrity, or availability of application data. The vulnerability is exploitable through remote input, as the function may process untrusted query strings supplied from outside the application.

The flaw exists in all releases of Sinaptik AI PandasAI up to and including version 3.0.0. Any installation that imports pandasai.helpers.sql_sanitizer and invokes is_sql_query_safe without additional safeguards is affected.

The CVSS score of 6.9 indicates a moderate severity. No EPSS score is available and the vulnerability is not listed in the CISA KEV catalog, but the availability of a public exploit and the remote trigger raise the likelihood of real‑world attacks. An attacker who can supply crafted query strings to the vulnerable function may execute path traversal attacks to read or modify files on the host system. The absence of an official vendor patch in the CVE data means the risk remains until a fix or workaround is applied.