Nearly 1 in 10 Exposed LiteLLM Gateways Accepted the Example "sk-1234" Admin Key

Summary

Nearly 10% of scanned LiteLLM servers accepted the default 'sk-1234' admin key. This key grants broad access, including API keys and cloud credentials. While LiteLLM considers some issues out-of-scope, several vulnerabilities have been fixed and some are being exploited.

Key Takeaways

1. Nearly one in ten of the internet-facing LiteLLM servers that Wiz Research scanned in February accepted sk-1234, the example admin key in LiteLLM's own setup guide.
2. Anyone who holds it can read every model provider's API key stored on the server. In Wiz's tests, it also reached the cloud IAM credentials of the machine the gateway runs on.
3. Upgrading to 1.84.0 or later covers every flaw in the table. The version ranges are as stated in the advisories.

Statement Breakdown

• Claimed Facts: 60% of statements the article presents as facts
• Opinions: 25% of statements classified as editorial or subjective
• Claims: 15% of statements surfaced for additional reader evaluation

Credibility & Bias Reasoning

Credibility assessment: The article presents technical details and references specific CVEs, indicating a degree of factual reporting. However, it relies heavily on a single research firm's findings and LiteLLM's own advisories, with some discrepancies in severity ratings. The lack of independent verification or broader industry consensus limits its overall credibility.

Bias assessment: Security Vulnerability Focus. The article's primary lens is the identification and detailing of security vulnerabilities within LiteLLM. It emphasizes potential risks and exploits, framing the subject matter through a security-first perspective. This focus, while informative, inherently highlights negative aspects and potential threats.

Note: This article details security vulnerabilities in LiteLLM. While informative, it relies on specific research and advisories; cross-referencing with official LiteLLM documentation and other security analyses is recommended for a comprehensive understanding.

Credibility flag: Technical, Security-Focused

Claimed Facts (8)

• This is a factual definition of LiteLLM's function and nature.
• This presents specific data points from Wiz Research's scan.
• This states a verifiable fact about the current state of LiteLLM's documentation.
• This describes a specific technical behavior of LiteLLM in earlier versions.
• This identifies a specific CVE and notes a difference in how it's described by different parties.
• This provides a timeline of when the reported flaws were addressed and disclosed.
• This states a fact about CISA's inclusion of a specific CVE in its catalog.
• This describes a specific incident reported by Microsoft involving LiteLLM exploitation.

Opinions (5)

• This interprets LiteLLM's security policy and frames it as a statement of their stance on certain vulnerabilities.
• This presents Wiz's interpretation of LiteLLM's design intent regarding a specific feature.
• The use of 'Disputed' indicates a subjective assessment of the differing descriptions of the CVE's severity.
• The word 'serious' conveys a judgment about the importance of the default value.
• The phrase 'a lot in reach' is a subjective assessment of the power an administrator holds.

Claims (6)

• This statement is presented without evidence or explanation for why a current figure cannot be determined, making it a potentially unsubstantiated claim.
• While likely true based on the context of the vulnerability, this is a technical assertion about the system's behavior that isn't directly proven within the text itself, and could be subject to nuanced interpretations.
• This is a strong technical assertion about the ineffectiveness of a security measure, presented without detailed proof within the article.
• The claim that 'no source reports' this is a negative assertion that is difficult to verify and could be incomplete. The statement that it 'needs admin access first' is a condition that might be true but is presented as a definitive barrier without further elaboration.
• This is a vague claim about LiteLLM's records not supporting a statement, lacking specific details or evidence to substantiate it.
• While likely true in context, this is a definitive statement about a system's credentialing mechanism that isn't explicitly demonstrated or proven within the provided text.

Key Sources

• The Hacker News — Media Outlet
• Wiz Research — Security Research Firm
• LiteLLM — Open-Source AI Gateway Project
• CISA — Cybersecurity and Infrastructure Security Agency
• Microsoft — Technology Company