# CVE-2026-34159

## Summary

- **CVE ID:** CVE-2026-34159
- **Severity:** CRITICAL
- **CVSS Score:** 9.8 (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H)
- **CWE:** CWE-119
- **Published:** Apr 1, 2026
- **Last Modified:** Apr 2, 2026

## Description

llama.cpp is an inference of several LLM models in C/C++. Prior to version b8492, the RPC backend's deserialize_tensor() skips all bounds validation when a tensor's buffer field is 0. An unauthenticated attacker can read and write arbitrary process memory via crafted GRAPH_COMPUTE messages. Combined with pointer leaks from ALLOC_BUFFER/BUFFER_GET_BASE, this gives full ASLR bypass and remote code execution. No authentication required, just TCP access to the RPC server port. This issue has been patched in version b8492.

## Affected Products

- ggml-org — llama.cpp (< b8492)

## References

- [CNA](https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-j8rj-fmpv-wcxw)
- [CNA](https://github.com/ggml-org/llama.cpp/pull/20908)
- [CNA](https://github.com/ggml-org/llama.cpp/commit/39bf0d3c6a95803e0f41aaba069ffbee26721042)

## Exploitation Prediction (EPSS)

- **EPSS Score:** 1.13%
- **EPSS Percentile:** 64.4

---
_Exported from OnDuty AI Vulnerability Intelligence on 2026-09-11._