CVE-2024-4888: LiteLLM arbitrary file deletion

CISO Take

If your team runs litellm as an LLM gateway or proxy, treat this as critical: any authenticated user can delete arbitrary server files—SSH keys, config files, SQLite databases—with a single crafted API call. Patch to the latest litellm release immediately, or block the /audio/transcriptions endpoint at the reverse proxy layer. No special skill is required; the attack is a one-liner.

What is the risk?

High risk (CVSS 8.1). litellm is a widely-deployed LLM proxy used to route traffic to OpenAI, Anthropic, and other providers. The vulnerability requires only low-privilege API credentials—typically any valid API key—with no user interaction and low attack complexity. Network-accessible by design, it offers attackers a trivially simple path to destructive impact (Integrity: High, Availability: High). Multi-tenant or shared litellm instances face compounded risk: one compromised API key can destroy shared infrastructure for all users.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
LiteLLM	pip	—	No patch
51.0K OpenSSF 6.1 6 dependents Pushed 3d ago 44% patched ~38d to patch Full package profile →

Do you use LiteLLM? You're affected.

How severe is it?

CVSS 3.1

8.1 / 10

EPSS

0.6%

chance of exploitation in 30 days

Higher than 45% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ CISA SSVC: Public PoC

○ Public PoC indexed (trickest/cve)

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AV Network

AC Low

PR Low

UI None

S Unchanged

C None

I High

A High

What should I do?

5 steps

Patch: Upgrade litellm to the latest release; validate the fix via the huntr advisory and changelog before deploying.
Network control: Restrict /audio/transcriptions to trusted internal IP ranges via reverse proxy (nginx/Caddy) as an immediate compensating control.
Least privilege: Run litellm as a non-root user with minimal filesystem permissions scoped to its working directory only.
Detection: Alert on unexpected file deletions in sensitive directories (/root/.ssh, /etc, app data dirs); monitor API logs for multipart/form-data requests to /audio/transcriptions with suspicious filename fields containing path separators.
Recovery readiness: Verify SSH keys and critical configs are backed up and recoverable before relying on patching alone.

What does CISA's SSVC say?

Decision Track*

Exploitation poc

Automatable No

Technical Impact partial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Auth Bypass DoS Framework API AML.T0037 - Data from Local System AML.T0040 - AI Model Inference API Access AML.T0049 - Exploit Public-Facing Application

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

ISO 42001

A.10.1 - Information security in AI system design

NIST AI RMF

MANAGE-2.2 - Mechanisms to sustain effectiveness of risk treatments

OWASP LLM Top 10

LLM04 - Model Denial of Service LLM08 - Excessive Agency

Frequently Asked Questions

What is CVE-2024-4888?

If your team runs litellm as an LLM gateway or proxy, treat this as critical: any authenticated user can delete arbitrary server files—SSH keys, config files, SQLite databases—with a single crafted API call. Patch to the latest litellm release immediately, or block the /audio/transcriptions endpoint at the reverse proxy layer. No special skill is required; the attack is a one-liner.

Is CVE-2024-4888 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2024-4888, increasing the risk of exploitation.

How to fix CVE-2024-4888?

1. Patch: Upgrade litellm to the latest release; validate the fix via the huntr advisory and changelog before deploying. 2. Network control: Restrict /audio/transcriptions to trusted internal IP ranges via reverse proxy (nginx/Caddy) as an immediate compensating control. 3. Least privilege: Run litellm as a non-root user with minimal filesystem permissions scoped to its working directory only. 4. Detection: Alert on unexpected file deletions in sensitive directories (/root/.ssh, /etc, app data dirs); monitor API logs for multipart/form-data requests to /audio/transcriptions with suspicious filename fields containing path separators. 5. Recovery readiness: Verify SSH keys and critical configs are backed up and recoverable before relying on patching alone.

What systems are affected by CVE-2024-4888?

This vulnerability affects the following AI/ML architecture patterns: LLM gateways and proxies, model serving, API gateways, agent frameworks.

What is the CVSS score for CVE-2024-4888?

CVE-2024-4888 has a CVSS v3.1 base score of 8.1 (HIGH). The EPSS exploitation probability is 0.61%.

What is the AI security impact?

Affected AI Architectures

LLM gateways and proxiesmodel servingAPI gatewaysagent frameworks

MITRE ATLAS Techniques

AML.T0037 Data from Local System

AML.T0040 AI Model Inference API Access

AML.T0049 Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.10.1

NIST AI RMF: MANAGE-2.2

OWASP LLM Top 10: LLM04, LLM08

What are the technical details?

Original Advisory

BerriAI's litellm, in its latest version, is vulnerable to arbitrary file deletion due to improper input validation on the `/audio/transcriptions` endpoint. An attacker can exploit this vulnerability by sending a specially crafted request that includes a file path to the server, which then deletes the specified file without proper authorization or validation. This vulnerability is present in the code where `os.remove(file.filename)` is used to delete a file, allowing any user to delete critical files on the server such as SSH keys, SQLite databases, or configuration files.

Exploitation Scenario

An attacker with a low-privilege litellm API key—obtained via credential stuffing, a leaked key in a public GitHub repository, or insider access—sends a POST request to /audio/transcriptions with a crafted multipart/form-data payload. Instead of a valid audio file, the filename field contains a path such as '../../.ssh/authorized_keys' or '/var/lib/litellm/litellm.db'. The server calls os.remove(file.filename) without any path validation or authorization check, silently deleting the targeted file. The attacker iterates through predictable high-value paths—SSH keys, the litellm database, environment files containing upstream API keys—causing service disruption, admin lockout, and destruction of audit trails, all while appearing as a legitimate API caller.

Weaknesses (CWE)

CWE-862 Missing Authorization CWE-862 Missing Authorization

CWE-862 — Missing Authorization: The product does not perform an authorization check when an actor attempts to access a resource or perform an action.

[Architecture and Design] Divide the product into anonymous, normal, privileged, and administrative areas. Reduce the attack surface by carefully mapping roles with data and functionality. Use role-based access control (RBAC) [REF-229] to enforce the roles at the appropriate boundaries. Note that this approach may not protect against horizontal authorization, i.e., it will not protect a user from attacking others with the same role.
[Architecture and Design] Ensure that access control checks are performed related to the business logic. These checks may be different than the access control checks that are applied to more generic resources such as files, connections, processes, memory, and database records. For example, a database may restrict access for medical records to a specific database user, but each record might only be intended to be accessible to the patient and the patient's doctor [REF-7].

Source: MITRE CWE corpus.