CVE-2026-12797: LiteLLM Has Security Feature Bypass in Guardrails

# CVE-2026-12797: LiteLLM Has Security Feature Bypass in Guardrails

https://gcdradar.vercel.app/incidents/cve-2026-12797-litellm-has-security-feature-bypass-in-guardr · AIR-2026-0783 · closed 2026-06-21

- Lane: Patch now
- Severity: 2 (Disclosed vulnerability)
- Fix status: Fix status unknown
- First reported: 2026-06-21
- AI role: Attack on AI
- Products: LiteLLM
- CVE ids: CVE-2026-12797
- CVSS: 6.3

## Summary

A security feature bypass vulnerability in LiteLLM affects safety guardrails due to call type mismatches. No active exploitation has been confirmed.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2026-06-21: CVE-2026-12797: A security flaw has been discovered in BerriAI litellm up to 1.82.5. (https://nvd.nist.gov/vuln/detail/CVE-2026-12797)
- 2026-06-21: CVE-2026-12797: BerriAI litellm has Security Feature Bypass in BannedKeywords and AzureContentSafety Guardrails via call_type Mismatch on Async Endpoints (https://github.com/advisories/GHSA-p897-vf7j-f5h8)

## Sources

- 2026-06-21 GitHub Advisory Database: CVE-2026-12797: BerriAI litellm has Security Feature Bypass in BannedKeywords and AzureContentSafety Guardrails via call_type Mismatch on Async Endpoints https://github.com/advisories/GHSA-p897-vf7j-f5h8
- 2026-06-21 NVD: CVE-2026-12797: A security flaw has been discovered in BerriAI litellm up to 1.82.5. https://nvd.nist.gov/vuln/detail/CVE-2026-12797

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.