●PocketClawAI security alerts
Sign in
highCVSS 8.8NVD status: Analyzed

CVE-2025-49847

llama.cpp is an inference of several LLM models in C/C++.

Componentllama.cpp
Severityhigh (CVSS 8.8)
CVSS vectorCVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Published (NVD)2025-06-17
Last modified2026-06-17
Affected versions
llama.cpp < b5662
Fixed inb5662
WeaknessCWE-119CWE-195
CISA KEVnot listed

NVD description

llama.cpp is an inference of several LLM models in C/C++. Prior to version b5662, an attacker‐supplied GGUF model vocabulary can trigger a buffer overflow in llama.cpp’s vocabulary‐loading code. Specifically, the helper _try_copy in llama.cpp/src/vocab.cpp: llama_vocab::impl::token_to_piece() casts a very large size_t token length into an int32_t, causing the length check (if (length < (int32_t)size)) to be bypassed. As a result, memcpy is still called with that oversized size, letting a malicious model overwrite memory beyond the intended buffer. This can lead to arbitrary memory corruption and potential code execution. This issue has been patched in version b5662.

Does CVE-2025-49847 hit the version you run?Paste your docker-compose.yml, requirements.txt or package.json. The scanner checks your pinned versions against this CVE and every other one we track for llama.cpp.
Check my stack

What to do

  • Upgrade llama.cpp to b5662 or later.
  • Do not expose the component's API or UI to the internet without authentication in front of it.

Sources

Other llama.cpp CVEs

high
CVE-2026-107183CVSS 8.1

llama.cpp before b11393 contains a use-after-free and double free vulnerability in common_chat_peg_mapper::map that allows unauthenticated r...

llama.cpp
2026-10-07medium
CVE-2026-86317CVSS 5.3

A vulnerability was detected in ggml-org llama.cpp up to 0.4.0. This impacts the function rpc_server::deserialize_tensor of the file ggml/sr...

llama.cpp
2026-09-07high
CVE-2026-52132CVSS 7.5

llama.cpp through commit 97f06e9, when started with the --reranking flag, allows remote attackers to cause a denial of service (std::bad_all...

llama.cpp
2026-09-01high
CVE-2026-52131CVSS 7.5

llama.cpp b5693 and before has a Reachable Assertion via the gguf_reader::read function.

llama.cpp
2026-09-01high
CVE-2026-52130CVSS 7.5

llama.cpp b5693 and before is vulnerable to Uncontrolled Recursion in common/json-schema-to-grammar.cpp, resulting in a denial of service.

llama.cpp
2026-09-01medium
CVE-2026-78148CVSS 5.3

A vulnerability was determined in ggml-org llama.cpp bec4772f6. This affects the function rpc_server::graph_compute of the file ggml/src/ggm...

llama.cpp
2026-08-24

All llama.cpp CVEs