A critical denial of service vulnerability in Ollama version 0.12.10 allows unauthenticated remote attackers to crash the service, potentially disrupting AI-driven workflows.

The vulnerability exists within the GGUF decoder component, where improper handling of input lengths leads to a panic state. This flaw is remotely exploitable by unauthenticated users, as indicated by the CVSS vector AV:N/PR:N.

The ability for an unauthenticated attacker to remotely crash the Ollama service poses a significant risk to operational continuity. Given the CVSS score of 7.5, this high-severity flaw could halt AI model inference and processing tasks, leading to service outages for applications dependent on the Ollama backend.

Immediate Action: Monitor the official Ollama GitHub repository and security advisories for the release of a patched version addressing this GGUF decoder issue.

Proactive Monitoring: Review system logs for frequent service restarts or "panic" errors associated with the Ollama process, which may indicate active exploitation attempts.

Compensating Controls: Implement network-level access controls or a Web Application Firewall to restrict access to the Ollama API to known, trusted IP addresses, thereby reducing the attack surface.

Public Exploit Available: Yes, a published proof-of-concept exists as documented in the security researcher's technical write-up referenced in the CVE record.

Organizations utilizing Ollama v0.12.10 should treat this vulnerability with high priority. While a formal patch status is currently unknown, administrators must limit network exposure of the service and prepare to deploy updates immediately upon vendor release to prevent potential service disruption.