CVE-2026-4996 PandasAI SQL 注入漏洞，影响版本

在 PandasAI 中发现了一个 SQL 注入漏洞，尤其是在 pandasai-lancedb 扩展程序版本 0.1.4 之前。此漏洞影响 lancedb.py 文件中的多个函数（具体来说是 deletequestionandanswers、deletedocs、updatequestionanswer、updatedocs、getrelevantquestionanswersbyid 和 getrelevantdocsbyid）。远程攻击者可以利用此漏洞来操纵 SQL 查询，从而可能访问、修改或删除 PandasAI 使用的 LanceDB 数据库中存储的敏感数据。漏洞利用程序的公开可用性加剧了情况，增加了攻击的风险。

Applications utilizing the PandasAI LanceDB Extension in versions 0.1.0 through 0.1.4 are at direct risk. This includes data science projects, AI applications, and any system relying on the extension for vector database interactions. Shared hosting environments where multiple applications share the same database are particularly vulnerable, as a compromise of one application could potentially lead to a compromise of the entire database.

• python / application: Inspect application logs for unusual SQL queries or error messages related to the vulnerable functions. Use a debugger to trace the execution flow and identify potential injection points. • generic web: Monitor web server access logs for requests containing suspicious SQL syntax or patterns targeting the affected endpoints. Use a WAF to detect and block malicious requests. • database (mysql, postgresql): Monitor database audit logs for unauthorized access attempts or suspicious SQL queries. Implement database activity monitoring (DAM) to detect and alert on anomalous behavior.

1. 2026-03-28Disclosure

   disclosure

1. 已保留2026-03-27
2. 发布日期2026-03-28
3. 修改日期2026-03-30
4. EPSS 更新日期2026-04-04

目前，PandasAI 开发人员尚未提供官方修复程序。最有效的即时缓解措施是升级到 0.1.4 之后的 PandasAI 版本，一旦可用。在此期间，建议限制对 PandasAI 实例和底层 LanceDB 数据库的访问。在应用程序代码中实施强大的输入验证和清理可以帮助减轻风险，尽管这需要彻底的代码审查。监控数据库活动以查找可疑模式也很重要。