CVE-2025-49847

{"cve": {"data_type": "CVE", "references": {"reference_data": [{"url": "https://github.com/ggml-org/llama.cpp/commit/3cfbbdb44e08fd19429fed6cc85b982a91f0efd5", "name": "https://github.com/ggml-org/llama.cpp/commit/3cfbbdb44e08fd19429fed6cc85b982a91f0efd5", "tags": [], "refsource": ""}, {"url": "https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-8wwf-w4qm-gpqr", "name": "https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-8wwf-w4qm-gpqr", "tags": [], "refsource": ""}]}, "data_format": "MITRE", "description": {"description_data": [{"lang": "en", "value": "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\u2019s 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."}]}, "problemtype": {"problemtype_data": [{"description": [{"lang": "en", "value": "CWE-119"}, {"lang": "en", "value": "CWE-195"}]}]}, "data_version": "4.0", "CVE_data_meta": {"ID": "CVE-2025-49847", "ASSIGNER": "security-advisories@github.com"}}, "impact": {}, "publishedDate": "2025-06-17T20:15Z", "configurations": {"nodes": [], "CVE_data_version": "4.0"}, "lastModifiedDate": "2025-06-17T20:50Z"}