Return to Threats

SearchLeak and LiteLLM CVE Chain Break Email and Key Trust Boundaries in GenAI Stacks

VentureBeat 2026-06-30 data leakage Critical

Executive Summary

VentureBeat details SearchLeak, where a copilot searched user mailboxes beyond intended scope, and a chain of three LiteLLM CVEs that leaked admin API keys.

What Happened

VentureBeat detailed two separate incidents where the SearchLeak tool and a chain of three LiteLLM CVEs led to sensitive data exposure and unauthorized access in AI application stacks.[11] In one case, a copilot was able to search user mailboxes beyond intended scope, and in another, LiteLLM misconfigurations and vulnerabilities resulted in leakage of admin API keys, highlighting data leakage and model-adjacent supply chain risks for SaaS and fintech teams building on LLM routing libraries.[11]

Affected Products

  • LiteLLM

Vulnerability / Attack Technique

Over-scoped copilot search; LiteLLM vulnerability chain

Technical Mechanism

The copilot's search reached mailboxes beyond its intended scope; LiteLLM flaws and misconfigurations exposed admin API keys.

AI Security Category

LLM02:2025 Sensitive Information Disclosure

Relevant Frameworks

LLM02:2025 Sensitive Information DisclosureLLM03:2025 Supply ChainLLM06:2025 Excessive Agency

RealGround Analysis

Both cases break trust boundaries around email and keys. RealGround analysis: scope copilot search permissions and patch LLM routing libraries.

Recommended Controls

  • Restrict copilot search scope to the user's own data
  • Patch LiteLLM and rotate admin keys

Related RealGround Service

AI Security Readiness Assessment · or start with the free AI Agent Security Scanner.

Sources

VentureBeat: https://venturebeat.com/security/searchleak-and-a-three-cve-litellm-chain-broke-the-same-ai-trust-boundary-in-two-weeks

Summary and analysis written by RealGround from the source above; last updated 2026-10-18.

Talk to AI CISO