Critical SQL Injection Vulnerability in LiteLLM Actively Exploited

Severity: High (Score: 72.8)

Sources: Bleepingcomputer, Cybersecuritynews, Gbhackers

Summary

Hackers are exploiting a critical SQL injection vulnerability in the LiteLLM open-source large-language model gateway, tracked as CVE-2026-42208. This pre-authentication flaw allows attackers to access sensitive information, including API keys and provider credentials, without authentication by sending a crafted Authorization header to any LLM API route. Exploitation began approximately 36 hours after the vulnerability was disclosed on April 24, 2026. The attackers have been observed executing targeted SQL injection attempts against specific database tables, indicating a high level of knowledge about the system's structure. LiteLLM, which has over 45,000 stars on GitHub, is widely used for managing AI models and has previously been targeted in a supply-chain attack. A patch was released in version 1.83.7 to mitigate this vulnerability, but instances running older versions are at risk. Security experts recommend rotating all credentials stored in exposed LiteLLM instances. For those unable to upgrade, a temporary workaround is available. Key Points: • CVE-2026-42208 is a critical SQL injection vulnerability in LiteLLM. • Attackers can exploit the flaw without authentication to access sensitive data. • A patch has been released, but many instances remain vulnerable.

Key Entities

• Sql Injection (attack_type)
• Supply Chain Attack (attack_type)
• LiteLLM (tool)
• CVE-2026-42208 (cve)
• CWE-287 - Improper Authentication (cwe)
• Cwe-89 - SQL Injection (cwe)
• T1190 - Exploit Public-Facing Application (mitre_attack)
• T1195 - Supply Chain Compromise (mitre_attack)
• PostgreSQL (platform)