VYPR
Moderate severityNVD Advisory· Published Aug 7, 2025· Updated Aug 7, 2025

CVE-2025-44779

CVE-2025-44779

Description

An issue in Ollama v0.1.33 allows attackers to delete arbitrary files via sending a crafted packet to the endpoint /api/pull.

AI Insight

LLM-synthesized narrative grounded in this CVE's description and references.

In Ollama v0.1.33, a crafted /api/pull request allows attackers to delete arbitrary files due to incorrect handling of digest file mismatches.

Vulnerability

Overview

CVE-2025-44779 affects Ollama versions <= 0.1.33. The vulnerability stems from an incorrect access control flaw in the /api/pull endpoint. When pulling a model, if a file already exists at the path where a digest is to be saved, the software treats it as having a mismatched digest value and deletes that file without proper authorization checks [2]. This allows an attacker to delete arbitrary files on the host system by sending a specially crafted packet to the endpoint.

Exploitation

Vector

To exploit this vulnerability, an attacker must be able to reach the Ollama HTTP API, typically listening on localhost:11434 by default, but potentially exposed to the network. The attacker crafts a POST request to /api/pull with parameters that force the server to check for a file at a chosen path. By providing a malicious manifest or digest reference, the server will attempt to save the digest and, upon finding a pre-existing file, delete it assuming the digest does not match [2]. The attack can be staged from a custom private registry server, as detailed in the advisory, making remote exploitation possible if the API is accessible [2].

Impact

Successful exploitation results in arbitrary file deletion. An attacker could delete critical system files, configuration files, or model data, leading to denial of service, data loss, or potential privilege escalation if deleted files affect security mechanisms. The vulnerability does not require authentication to the Ollama service [2].

Mitigation

The vulnerability has been patched in Ollama version 0.1.34 and later. Users are strongly advised to upgrade immediately. There is no known workaround; restricting network access to the Ollama API can reduce risk but does not fully mitigate the flaw [2]. This CVE is not currently listed in CISA's Known Exploited Vulnerabilities catalog.

AI Insight generated on May 19, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.

Affected packages

Versions sourced from the GitHub Security Advisory.

PackageAffected versionsPatched versions
github.com/ollama/ollamaGo
< 0.1.340.1.34

Affected products

2
  • Ollama/Ollamadescription
  • Ollama/Ollamallm-fuzzy
    Range: =0.1.33

Patches

0

No patches discovered yet.

Vulnerability mechanics

AI mechanics synthesis has not run for this CVE yet.

References

6

News mentions

0

No linked articles in our index yet.