Ollama Out-of-Bounds Read Vulnerability Allows Remote Process Memory Leak
A serious security flaw in Ollama has been identified by cybersecurity researchers. If this flaw is successfully exploited, a remote, unauthenticated attacker may be able to access the entire process memory.
Tracked as CVE-2026-7482 (CVSS score: 9.1), the out-of-bounds read vulnerability probably affects more than 300,000 servers worldwide. Cyera has given it the code name Bleeding Llama.
Large language models (LLMs) can be executed locally rather than in the cloud thanks to the well-known open-source framework Ollama. The project has received over 171,000 stars on GitHub and has been forked more than 16,100 times.
A description of the vulnerability at CVE.org states that Ollama prior to 0.17.1 has a heap out-of-bounds read vulnerability in the GGUF model loader. An attacker-supp...