CLEANSTART-2026-TE29052 Security fixes in litellm-database 1.97.0-r0

Package litellm-database version 1.97.0-r0 fixes 5 vulnerabilities: CVE-2026-54282, CVE-2025-62727, CVE-2026-48818, CVE-2026-54283, CVE-2025-54121...

CLEANSTART-2026-AL63130 Security fixes in litellm-database 1.96.0-r0

Package litellm-database version 1.96.0-r0 fixes 5 vulnerabilities: CVE-2026-54282, CVE-2025-62727, CVE-2026-48818, CVE-2026-54283, CVE-2025-54121...

CLEANSTART-2026-CF86587 Security fixes in litellm-database 1.94.3-r0

Package litellm-database version 1.94.3-r0 fixes 3 vulnerabilities: CVE-2026-69247, CVE-2026-69248, CVE-2026-69249...

Arbitrary Code Injection

Overview litellm is a Library to easily interface with LLM API providers Affected versions of this package are vulnerable to Arbitrary Code Injection in the /prompts/test endpoint through the renderedcontent path in litellm/proxy/prompts/promptendpoints.py. An attacker can execute arbitrary OS...

GHSA-6WVF-77M9-58RM LiteLLM vulnerable to server-side template injection in the /prompts/test endpoint

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

LiteLLM vulnerable to server-side template injection in the /prompts/test endpoint

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

CVE-2026-37004

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

Inside 90 days of attacks on AI infrastructure

Wiz honeypots uncover active campaigns targeting LiteLLM, MCP servers, and AI frameworks through RCE, blind prompt injection, and memory credential theft...

Alleged TeamPCP Hackers Charged in Australia Over Major Supply Chain Attacks

The Australian Federal Police AFP has charged two Western Australian men with a combined total of 14 offences over their alleged role in TeamPCP , the cybercrime group behind the March 2026 compromise of the open-source security scanners Trivy and Checkmarx KICS and the AI gateway LiteLLM. Louis...

CVE-2026-37004

BerriAI litellm versions

CVE-2026-37004

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

CVE-2026-37004

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

CVE-2026-37004

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

EUVD-2026-67271

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

PT-2026-82662

Name of the Vulnerable Software and Affected Versions litellm versions prior to 1.82.5 Description An issue exists where unauthenticated remote attackers can execute arbitrary OS commands. This occurs due to the use of an unsandboxed jinja2.Environment, which enables Server-Side Template Injectio...

CVE-2026-37004: Improper Neutralization of Special Elements Used in a Template Engine

BerriAI litellm =1.82.4 is vulnerable to Server-Side Template Injection SSTI, which allows unauthenticated remote attackers to execute arbitrary OS commands via a crafted dotpromptcontent parameter in the /prompts/test endpoint due to use of an unsandboxed jinja2.Environment...

When AI infrastructure becomes the target: Securing gateways and control points

In this article 1. AI workloads are becoming high-value control points 2. Case study 1: LiteLLM gateway compromise 3. Case study 2: RAGFlow compromise 4. Case study 3: Kestra compromise 5. Mitigation and protection guidance 6. MITRE ATT&CK techniques observed 7. References 8. Learn more AI is...

When AI infrastructure becomes the target: Securing gateways and control points

In this article 1. AI workloads are becoming high-value control points 2. Case study 1: LiteLLM gateway compromise 3. Case study 2: RAGFlow compromise 4. Case study 3: Kestra compromise 5. Mitigation and protection guidance 6. MITRE ATT&CK techniques observed 7. References 8. Learn more AI is...

CVE-2026-12795

A flaw was found in BerriAI litellm. This vulnerability allows a remote attacker to bypass authentication within the SSO Debug Flow component by manipulating the json.dumps function. The successful exploitation of this flaw could lead to unauthorized access and potential information disclosure...

Malicious LiteLLM Releases Tied to Trivy Hack May Have Exposed 2,100+ Organizations

Two malicious LiteLLM releases sat on PyPI for about 40 minutes in March carrying credential-stealing code capable of harvesting cloud keys, SSH keys, Kubernetes tokens, database passwords, and other secrets from systems that installed them. Threat intelligence firm CloudSEK now says a dataset it...