CVE-2025-0317
📋 TL;DR
A vulnerability in Ollama versions up to 0.3.14 allows attackers to upload malicious GGUF model files that trigger a division by zero error, causing server crashes and denial of service. This affects all Ollama servers running vulnerable versions, particularly those exposed to untrusted users who can upload models.
💻 Affected Systems
- ollama/ollama
📦 What is this software?
Ollama by Ollama
⚠️ Risk & Real-World Impact
Worst Case
Sustained DoS attacks could render Ollama services completely unavailable, disrupting AI/ML workflows and dependent applications.
Likely Case
Intermittent server crashes requiring manual restarts, causing service disruption and potential data loss for in-progress operations.
If Mitigated
Limited impact with proper access controls and monitoring, allowing quick detection and recovery from attacks.
🎯 Exploit Status
Requires ability to upload GGUF model files to the Ollama server, which typically requires some level of access or API usage.
🛠️ Fix & Mitigation
✅ Official Fix
Patch Version: > 0.3.14
Vendor Advisory: https://huntr.com/bounties/a9951bca-9bd8-49b2-b143-4cd4219f9fa0
Restart Required: No
Instructions:
1. Update Ollama to version 0.3.15 or later. 2. Use package manager: 'ollama upgrade' or download latest release from GitHub. 3. No restart required as updates take effect immediately.
🔧 Temporary Workarounds
Restrict model uploads
allLimit who can upload GGUF models to trusted users only using authentication and authorization controls.
Network segmentation
allPlace Ollama servers behind firewalls and restrict access to trusted IP addresses only.
🧯 If You Can't Patch
- Implement strict access controls to limit who can upload models to the Ollama server
- Deploy monitoring and alerting for server crashes with automated restart capabilities
🔍 How to Verify
Check if Vulnerable:
Run 'ollama --version' and check if version is 0.3.14 or earlier.Check Version:
ollama --versionVerify Fix Applied:
After update, run 'ollama --version' to confirm version is 0.3.15 or later.📡 Detection & Monitoring
Log Indicators:
- Server crash logs mentioning 'division by zero' or 'ggufPadding'
- Unexpected process termination of ollama service
- Multiple failed model upload attempts
Network Indicators:
- Unusual spikes in model upload requests
- Repeated connections followed by service unavailability
SIEM Query:
process.name:"ollama" AND (event.action:"crashed" OR log.message:"*division by zero*" OR log.message:"*ggufPadding*")