{"answers":{"affected":"It affects 1 product.","exploited":"It is not confirmed as exploited in the wild, but public exploit code for CVE-2025-49847 exists.","fix":"Upgrade to a fixed version.","what":"CVE-2025-49847 is a high-severity memory corruption vulnerability in ggml-org llama.cpp, published on 17 Jun 2025. An attacker can exploit it over the internet without logging in, and only if a user opens a file or follows a link."},"fix_prompt":"Check this codebase for CVE-2025-49847 and fix it if it is affected.\n\nWhat it is: CVE-2025-49847 is a high-severity memory corruption vulnerability in ggml-org llama.cpp, published on 17 Jun 2025. An attacker can exploit it over the internet without logging in, and only if a user opens a file or follows a link.\nExploitation: It is not confirmed as exploited in the wild, but public exploit code for CVE-2025-49847 exists.\n\nAffected product: ggml-org llama.cpp. No package-level version data is available, so check the vendor advisory linked from the reference page.\n\nTask:\n1. Search the dependency manifests and lockfiles (for example package.json, package-lock.json, requirements.txt, poetry.lock, uv.lock, pom.xml, build.gradle, go.mod, Cargo.lock, Gemfile.lock) and any Dockerfiles for the affected packages, including transitive dependencies.\n2. If an affected version is in use, upgrade it to the fixed version for its release line with the project's package manager, so the lockfile is regenerated. Prefer the smallest upgrade that fixes it.\n3. If there is no fix, or the upgrade would break the build, do not guess: explain the risk and the safest workaround instead.\n4. Run the tests and summarise what changed and why.\n5. If nothing in the project is affected, say so and change nothing.\n\nReference: https://insights.sbomify.com/CVE-2025-49847/ (Markdown: https://insights.sbomify.com/CVE-2025-49847/index.md)","id":"CVE-2025-49847","markdown":"https://insights.sbomify.com/CVE-2025-49847/index.md","name":"CVE-2025-49847","record":{"aliases":["DEBIAN-CVE-2025-49847","GHSA-8wwf-w4qm-gpqr","UBUNTU-CVE-2025-49847","openSUSE-SU-2025:15245-1"],"analysis":{"component_type":"library","fix_type":"upgrade","impacts":["code_execution"],"network_reachable":"yes","non_default_config":false,"product_category":"LLM inference library","requires_auth":"none","summary":"An attacker can achieve arbitrary code execution by supplying a malicious GGUF model that triggers a buffer overflow in llama.cpp's vocabulary loading.","user_interaction":"required","vuln_class":"memory_corruption"},"analysis_generated_by":"qwen3.8-flash-next","assigner":"GitHub_M","cpe":[{"product":"llama.cpp","vendor":"ggml"}],"cvss":{"score":8.8,"severity":"high","source":"cna","vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H","version":"3.1"},"cwes":[{"id":"CWE-195","name":"Signed to Unsigned Conversion Error"},{"id":"CWE-119","name":"Improper Restriction of Operations within the Bounds of a Memory Buffer"}],"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 \u003c (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.","ecosystems":["openSUSE:Tumbleweed"],"exploits":{"exploitdb":0,"first_seen":null,"github_poc":0,"max_poc_stars":null,"metasploit":0,"nuclei":0,"refs":[],"tier":"weaponised"},"has_fix":true,"headline":"llama.cpp Vulnerable to Buffer Overflow via Malicious GGUF Model","id":"CVE-2025-49847","indexable":true,"kev":null,"kind":"cve","links":{"ecosystem":null,"product":null,"products":[],"type":"memory_corruption","year":"2025"},"n_packages":1,"n_references":2,"next":{"headline":"Out-of-bounds Write in LS Electric GMWin 4","id":"CVE-2025-49848"},"packages":[{"ecosystem":"openSUSE:Tumbleweed","fixed":["5699-1.1"],"introduced":["0"],"last_affected":[],"package":"llamacpp","purl":"pkg:rpm/opensuse/llamacpp\u0026distro=openSUSE%20Tumbleweed","source":"openSUSE-SU-2025:15245-1"}],"prev":{"headline":"wire-ios accidentally logs message contents","id":"CVE-2025-49846"},"products":["llama.cpp"],"published":"2025-06-17","related":[{"cvss_score":10,"headline":"llama.cpp allows write-what-where in rpc_server::set_tensor","id":"CVE-2024-42479","kev_date_added":null,"published":"2024-08-12"},{"cvss_score":9.8,"headline":"llama.cpp: Unauthenticated RCE via GRAPH_COMPUTE buffer=0 bypass in llama.cpp RPC backend","id":"CVE-2026-34159","kev_date_added":null,"published":"2026-04-01"},{"cvss_score":9.2,"headline":"llama.cpp b4882–b9058 Buffer Overflow in KV Cache State Restore","id":"CVE-2026-43629","kev_date_added":null,"published":"2026-08-06"},{"cvss_score":9.2,"headline":"llama.cpp b7492–b9060 Use-After-Free RCE via llama-server","id":"CVE-2026-43631","kev_date_added":null,"published":"2026-08-06"},{"cvss_score":9.2,"headline":"llama.cpp b7492–b9060 Use-After-Free in Tokenization Endpoints","id":"CVE-2026-43632","kev_date_added":null,"published":"2026-08-06"},{"cvss_score":8.8,"headline":"llama.cpp has Out-of-bounds Write in llama-server","id":"CVE-2026-21869","kev_date_added":null,"published":"2026-01-07"}],"sources":[{"class":"attribution","licence":"CVE Terms of Use (MITRE)","name":"CVE List","prefix":"CVE"},{"class":"attribution","licence":"CC BY 4.0","name":"SUSE","prefix":"openSUSE"},{"class":"attribution","licence":"US public information, NVD API terms","name":"NVD"},{"class":"attribution","licence":"CWE Terms of Use","name":"CWE (MITRE)"},{"class":"attribution","licence":"FIRST CVSS licence","name":"CVSS (FIRST)"}],"ssvc":{"automatable":"no","exploitation":"poc","technical_impact":"total"},"state":"PUBLISHED","updated":"2025-06-18","vendors":["ggml-org"]},"summary":"An attacker can achieve arbitrary code execution by supplying a malicious GGUF model that triggers a buffer overflow in llama.cpp's vocabulary loading.","url":"https://insights.sbomify.com/CVE-2025-49847/"}