Actively exploited vulnerabilities affecting BerriAI products.https://wildfortech.comhttps://wildfortech.com/security#CVE-2026-42271https://wildfortech.com/security#CVE-2026-42271This vulnerability in BerriAI LiteLLM allows any authenticated user — even those with low-privilege internal-user keys — to inject and execute arbitrary commands directly on the underlying host system. This is a critical risk because it means attackers don't need administrative access to take control of the server. A compromised or malicious low-privilege account could be used to exfiltrate data, pivot to other systems, or fully compromise the host running LiteLLM.Mon, 08 Jun 2026 00:00:00 GMTBerriAILiteLLMhttps://wildfortech.com/security#CVE-2026-42208https://wildfortech.com/security#CVE-2026-42208BerriAI LiteLLM, a popular proxy for managing LLM API calls, contains a SQL injection flaw that lets an attacker read and potentially modify data in the proxy's underlying database. Because LiteLLM stores credentials for AI services, a successful exploit could expose API keys and access tokens for multiple LLM providers, leading to unauthorized use of those services and potential data exfiltration from any system the proxy touches.Fri, 08 May 2026 00:00:00 GMTBerriAILiteLLM