CVE-2026-33298Disclosure(ggml / llama.cpp)

MEDIUMCVSS 7.8 · HIGH

Exploitation ongoing with high activity in latest observed window (1 mentions)

Immediate actions

Prioritize remediation for ggml llama.cpp systems immediately
Assume compromise if assets are exposed
Hunt for exploitation attempts and persistence artifacts
Increase monitoring for publicly documented tradecraft
Track advisory updates for patch or workaround availability

Recommended action window: Immediate (within 24h)

NVD description

llama.cpp is an inference of several LLM models in C/C++. Prior to b7824, an integer overflow vulnerability in the `ggml_nbytes` function allows an attacker to bypass memory validation by crafting a GGUF file with specific tensor dimensions. This causes `ggml_nbytes` to return a significantly smaller size than required (e.g., 4MB instead of Exabytes), leading to a heap-based buffer overflow when the application subsequently processes the tensor. This vulnerability allows potential Remote Code Execution (RCE) via memory corruption. b7824 contains a fix.

ggml CVEs Back to dashboard

5.0/ 10 priority

Sources & remediation

Exploit / PoC references

https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-96jg-mvhq-q7q7

Other references

https://github.com/ggml-org/llama.cpp/releases/tag/b7824

Weakness type (CWE)

CWE-122CWE-190

Priority

MEDIUM

Exploitation

ACTIVE

PoC

YES

Patch

NONE

Momentum

STABLE

Are you affected?

If you run products in this scope, you should treat this CVE as relevant to your environment.

llama.cpp

Threat summary

Active exploitation appears in 1 classified signals
Public PoC is present in monitored signal
3 mentions across 3 observed days
Momentum state: stable

What's happening

Active exploitation reported across 1 signal
PoC mentioned or linked in 1 signal
Technical details provided in 1 signal
Disclosure: 1 classified signal
General: 1 classified signal
Peaked 2d ago at 1 mentions (2026-03-24); latest day: 1
3 total mentions across 3 days

Affected systems

Vendors

ggml

Products

llama.cpp

Deep dive

Activity timeline3 mentions / 3d

Signal classification3 categories

Disclosure

133.3%

General

133.3%

Active Exploitation

133.3%

Referenced assets3 URLs

By indicator

Classification over time

Date	Total	Labels

2026-03-24	1	Disclosure1
2026-03-25	1	General1
2026-03-29	1	Active Exploitation1

Full discourse3 posts

VulDB 🛡@vuldb194d
Active Exploitation
Our CTI team identified a lot of activities targeting ggml-org llama.cpp (CVE-2026-33298) https://vuldb.com/vuln/352585/cti
Post summary
CTI analysts report multiple attacks targeting CVE‑2026‑33298 and reference a Vuldb entry, indicating active exploitation in the wild.
02051178
2.1K followersView on X
CVEarity@CVEarity198d
General
⚡ New CVE Alert: CVE-2026-33298 📊 Severity: 7.8 🚨 Risk Level: High 🧩 Affects: Multiple / Unspecified Products Reference: https://nvd.nist.gov/vuln/detail/CVE-2026-33298 #CVE-2026-33298 #CVE #High #CyberSecurity #InfoSec https://t.co/EWUy1RBQ3M
Post summary
The tweet announces CVE-2026-33298 with a 7.8 severity rating and high risk but provides no further technical or operational details.
0000035
114 followersView on X
CVE@CVEnew199d
Disclosure
CVE-2026-33298 llama.cpp is an inference of several LLM models in C/C++. Prior to b7824, an integer overflow vulnerability in the `ggml_nbytes` function allows an attacker to bypass… https://www.cve.org/CVERecord?id=CVE-2026-33298
Post summary
A newly disclosed integer overflow vulnerability in llama.cpp’s ggml_nbytes function (CVE‑2026‑33298) is highlighted without any PoC, exploit, or patch details.
00000145
56.8K followersView on X

CPE platform detail1 entries

Part:

Vendor:

Product:

Version:

Target SW:

Target HW:

1 of 1 entries

Part	Vendor	Product	Version	Target SW	Target HW
App	ggml	llama.cpp	-	-	-

Explore more

Back to dashboard ggml CVEs llama.cpp vulnerabilities