A critical denial of service vulnerability exists in Ollama that allows unauthenticated remote attackers to crash the application.

This is a denial of service vulnerability located in the readGGUFV1String function within the fs/ggml/gguf.go file. The function fails to safely handle string lengths provided in untrusted GGUF metadata, allowing an unauthenticated attacker to trigger a process panic.

The ability for an unauthenticated remote attacker to crash the service results in significant operational downtime and service unavailability. Given the CVSS score of 7.5, this high severity flaw poses a substantial risk to business continuity, particularly for organizations relying on Ollama for AI model inference tasks.

Immediate Action: Monitor the official Ollama GitHub repository for the release of a patched version and apply the update immediately upon availability.

Proactive Monitoring: Review system and application logs for unexpected crashes or error patterns related to GGUF metadata processing.

Compensating Controls: Implement network-level access controls or a Web Application Firewall (WAF) to restrict traffic to the Ollama service to trusted sources only, thereby reducing the attack surface.

Public Exploit Available: Yes, a published proof-of-concept exists (attributed to the research write-up at zero.shotlearni.ng).

The presence of a public proof-of-concept for this denial of service vulnerability necessitates immediate vigilance. Administrators should track the Ollama project for official remediation and apply the fix as soon as it is released to prevent potential service disruption by remote actors.