A critical authentication bypass vulnerability, CVE-2026-59822, has been discovered in LiteLLM versions prior to 1.84.0. This flaw allows unauthenticated attackers to exploit the MCP Streamable HTTP endpoint, enabling them to list and invoke configured MCP tools. The vulnerability is particularly serious because LiteLLM gateways often serve as a central hub for API keys and access to sensitive internal resources, making them a prime target for attackers. Security researchers have linked this bypass, along with other vulnerabilities, to sophisticated campaigns by groups like Qilin ransomware, highlighting the growing threat to self-hosted AI infrastructure. AI
IMPACT Operators must upgrade LiteLLM to patch a critical vulnerability that exposes AI gateways to unauthorized access and potential compromise.
RANK_REASON Security vulnerability disclosure affecting an AI infrastructure tool.
- Cisa
- CVE-2026-42271
- CVE-2026-48710
- CVE-2026-59822
- CVSS 3.1
- CWE-287
- GHSA-7488-6r32-c95q
- LiteLLM
- LiteLLM 1.84.0
- MCP
- Microsoft
- Qilin
- Starlette
- Wiz Lighting
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →