Description
The Large Language Model (LLM) has disclosed a list of internal tools and their descriptions upon receiving specially crafted queries. This information exposure can aid attackers in reconnaissance, potentially enabling more targeted attacks or further vulnerability enumeration.
Remediation
Configure the LLM to restrict or sanitize responses regarding its internal tools and capabilities. Implement validation checks or filters to detect and prevent sensitive information disclosure.
References
Related Vulnerabilities
MediaWiki Exposure of Sensitive Information to an Unauthorized Actor Vulnerability (CVE-2017-0361)
MongoDb Exposure of Sensitive Information to an Unauthorized Actor Vulnerability (CVE-2016-6494)
Oracle JRE Exposure of Sensitive Information to an Unauthorized Actor Vulnerability (CVE-2024-21147)
WordPress 4.8.x Multiple Vulnerabilities (4.8 - 4.8.15)
Lighttpd Exposure of Sensitive Information to an Unauthorized Actor Vulnerability (CVE-2008-4360)