Security Resilience

CVE-2026-70640

Memory Safety in Ggml Llama.Cpp b1886 – b7445

Published
06 August 2026
Modified
29 October 2026
Patch / advisory
CVSS Score v4 7.3
Click a component to see what it means
Raw vectorCVSS:4.0/AV:L/AC:H/AT:N/PR:N/UI:P/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
EPSS Score 0.0024 15th percentile
Risk Priority 55 floored blend · peak EPSS

CVSS and EPSS are reproduced from their sources (NVD, FIRST EPSS). EPSS, Forum of Incident Response and Security Teams (FIRST), epss.first.org. Risk Priority is our own derived reading, not an NVD score.

Summary

CVE-2026-70640 is a high-severity Race Condition (CWE-362) vulnerability in Ggml Llama.Cpp. Its CVSS base score is 7.3 (High).

Operationally, exploitation aligns with the MITRE ATT&CK technique Exploitation for Privilege Escalation (T1068); ranked at the 15th percentile by exploit likelihood (below the median); it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.

This vulnerability is AI-related — categorised as NLP and Transformers.

The strongest mitigations our analysis identified map to SA-11 (Developer Testing and Evaluation) and SA-15 (Development Process, Standards, and Tools) — see the control section below for these in your framework.

OWASP Top 10 for Web (2025)

EU & UK References

Vulnerability Data

llama.cpp builds b1886 through b7445 contain a race condition use-after-free vulnerability in the LLaMA-Android JNI wrapper where bench_1model() and free_1context() lack synchronization, allowing Thread A to operate on freed memory while Thread B concurrently frees the llama_context. Attackers can exploit…

more

this by performing heap spray with attacker-controlled data containing a fake vtable to hijack the vtable pointer at offset +0x30, causing llama_batch_allocr::clear() to dereference arbitrary memory and achieve remote code execution.

CWE(s)

AI Security AnalysisAI

AI Category
NLP and Transformers
Risk Domain
N/A
OWASP Top 10 for LLMs 2025
None mapped
Classification Reason
Matched keywords: llama.cpp, llama

Related Threats

MITRE ATT&CK Enterprise Techniques

MITRE ATT&CK v20.0

T1068 Exploitation for Privilege Escalation Privilege Escalation
Adversaries may exploit software vulnerabilities in an attempt to elevate privileges.
T1190 Exploit Public-Facing Application Initial Access
Adversaries may attempt to exploit a weakness in an Internet-facing host or system to initially access a network.
T1203 Exploitation for Client Execution Execution
Adversaries may exploit software vulnerabilities in client applications to execute code.
T1499 Endpoint Denial of Service Impact
Adversaries may perform Endpoint Denial of Service (DoS) attacks to degrade or block the availability of services to users.
T1499.004 Application or System Exploitation Impact
Adversaries may exploit software vulnerabilities that can cause an application or system to crash and deny availability to users.
T1687 Exploitation for Defense Impairment Defense Impairment
Adversaries may exploit vulnerabilities in security software, infrastructure, or defensive components to degrade, disable, or otherwise continue to impair their ability to prevent, detect, or respond to malicious activity.
Derived from this CVE’s CWE(s) via the direct CWE→ATT&CK cross-walk.

CVEs Like This One

CVE-2024-41130Same product: Ggml Llama.Cpp
CVE-2026-43631Same product: Ggml Llama.Cpp
CVE-2024-24855Shared CWE-362, CWE-476
CVE-2023-31083Shared CWE-362, CWE-476
CVE-2024-43866Shared CWE-362, CWE-476
CVE-2026-53242Shared CWE-362, CWE-476
CVE-2026-23202Shared CWE-362, CWE-476
CVE-2024-56568Shared CWE-362, CWE-476
CVE-2024-35977Shared CWE-362, CWE-476
CVE-2023-2898Shared CWE-362, CWE-476

Affected Assets

ggml
llama.cpp
b1886 — b7445

Mitigating Controls

Mitigating Controls (NIST 800-53 r5)

Our mappings, made for this CVE.

Developer testing and evaluation (including static analysis) directly finds null-dereference bugs before deployment.

Documented development standards and tools can enforce null-safety rules and safe pointer usage.

Engineering principles can mandate defensive coding such as explicit null checks before dereference.

Maintaining separate execution domains for each process structurally eliminates unintended concurrent access to the same shared resources.

Preventing unintended information transfer through shared system resources directly addresses the improper concurrent modification that defines a race condition.

Mitigating Controls (NIST CSF 2.0)

Our mappings, derived directly from the weakness types (CWEs) cited in the NVD entry via our CWE→CSF cross-walk — links open the control.

PR.PS-06 mostly match
blocks

Secure SDLC practices (static analysis, code review, safe coding standards) directly prevent NULL dereference bugs during development.

Mitigating Controls (ISO/IEC 27001:2022 Annex A)

Our mappings, derived directly from the weakness types (CWEs) cited in the NVD entry via our CWE→ISO cross-walk — links open the control.

finds

Security testing can detect race conditions, but does not prevent them at design or coding time.

prevents

Secure coding standards explicitly forbid unsafe concurrent access patterns and mandate atomic operations or locks.

Hardening callouts derived

DISA STIG baselines are authored against specific weaknesses. Where this CVE cites a broad weakness class, the mapping is not asserted at that abstraction (owner ruling, 2026-09-25: grade only at Base and below), so no rule is shown here rather than a rule shown against the wrong grain.

  • No fine-grained STIG row for CWE-362, a class-level weakness, and none of its children carries one either.

References