CVE-2026-35029

Multiple critical vulnerabilities were discovered in LiteLLM, a popular open-source LLM gateway, enabling authentication bypass, remote code execution, and cloud credential theft. CVE-2026-59822 allows unauthenticated access to MCP endpoints via a Bearer token bypass, while CVE-2026-59821 enables post-authentication root-level RCE through unsandboxed custom code guardrails. A default master key (sk-1234) is accepted by 9.6% of public instances, allowing attackers to achieve admin access and exploit these vulnerabilities. Additionally, pass-through endpoints can be abused to exfiltrate cloud metadata and IAM credentials, especially when combined with default or missing authentication.

A critical vulnerability named AgentForger in OpenAI's ChatGPT Workspace Agents could allow attackers to deploy rogue AI agents via a phishing link. The flaw, a cross-site request forgery (CSRF), enables automatic creation and execution of malicious agents within an authenticated user's session without further interaction. These agents can persist, execute tasks from emails, access enterprise data, and send phishing messages, effectively becoming autonomous insiders.