Skip to main content

NVIDIA Megatron Bridge EUVDEUVD-2026-41019

| CVE-2026-24250 HIGH
Deserialization of Untrusted Data (CWE-502)
2026-07-01 nvidia GHSA-qhf9-p9vv-vgj2
7.8
CVSS 3.1 · Vendor: nvidia
Share

Severity by source

Vendor (nvidia) PRIMARY
7.8 HIGH
AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
vuln.today AI
7.8 HIGH

Deserialization is reachable only locally with existing low privileges (AV:L/PR:L), triggers without interaction (AC:L/UI:N), and yields full code execution giving high C/I/A within an unchanged scope.

3.1 AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
4.0 AV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N

Primary rating from Vendor (nvidia).

CVSS VectorVendor: nvidia

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Attack Vector
Local
Attack Complexity
Low
Privileges Required
Low
User Interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

Lifecycle Timeline

2
Analysis Generated
Jul 01, 2026 - 15:54 vuln.today
CVE Published
Jul 01, 2026 - 14:58 cve.org
HIGH 7.8

DescriptionCVE.org

NVIDIA Megatron Bridge for Linux contains a vulnerability where an attacker could cause improper validation of allowed inputs. A successful exploit of this vulnerability might lead to code execution, escalation of privileges, data tampering, and information disclosure.

AnalysisAI

Local privilege escalation and code execution in NVIDIA Megatron Bridge for Linux stems from unsafe deserialization of attacker-controlled input (CWE-502), allowing a low-privileged local user to achieve arbitrary code execution, tamper with data, and disclose information. NVIDIA reported the flaw with no public exploit identified at time of analysis, and it is not listed in CISA KEV; no EPSS score was provided. Megatron Bridge is an ML/LLM training framework, so impact centers on shared GPU/training hosts rather than internet-facing services.

Technical ContextAI

Megatron Bridge is NVIDIA's bridging library within the Megatron ecosystem, used to connect model/checkpoint and configuration state for large-language-model training on Linux GPU hosts. The root cause is CWE-502 (Deserialization of Untrusted Data): the code deserializes inputs - typically model checkpoints, configuration blobs, or serialized Python objects (e.g. pickle-backed artifacts common in ML tooling) - without adequately validating that the input is trusted or well-formed. Because Python deserialization can instantiate arbitrary objects and invoke callables during unpickling, an attacker who supplies a crafted artifact can direct execution flow, which the vendor maps to code execution, privilege escalation, data tampering, and information disclosure. The affected component is identified by CPE cpe:2.3:a:nvidia:megatron-bridge:*:*:*:*:*:*:*:* (all versions until the fixed release).

RemediationAI

Patch available per vendor advisory: upgrade NVIDIA Megatron Bridge to the fixed release referenced in NVIDIA's product-security advisory at https://github.com/NVIDIA/product-security/tree/main/2026/5841 (consult that advisory for the exact patched version, which was not included in this dataset). Until patched, restrict who can supply model checkpoints, configuration files, and other serialized artifacts to Megatron Bridge, and load artifacts only from trusted, integrity-verified sources - treat third-party or downloaded checkpoints as untrusted, since loading them is the exploitation path and blocking untrusted artifacts is the most effective compensating control, at the cost of workflow friction. Run training jobs under least-privilege accounts and isolate them (dedicated users, containers, or namespaces) so that successful code execution cannot escalate beyond the training context, accepting added operational overhead. Where feasible, prefer safe serialization formats (e.g. safetensors) over pickle-based deserialization for model weights to shrink the attack surface.

CVE-2026-24248 HIGH
7.8 Jul 01

Arbitrary code execution in NVIDIA Megatron Bridge for Linux stems from improper control of code generation (CWE-94), al

CVE-2026-24243 HIGH
7.8 Jul 01

Arbitrary code execution in NVIDIA Megatron Bridge (all versions per the NVIDIA advisory) arises from unsafe deserializa

CVE-2026-24247 HIGH
7.8 Jul 01

Insecure deserialization in NVIDIA Megatron Bridge for Linux (CWE-502) lets an attacker who supplies a crafted serialize

CVE-2026-24246 HIGH
7.8 Jul 01

Arbitrary code execution in NVIDIA Megatron Bridge on Linux arises from unsafe reflection (CWE-470), where externally-co

CVE-2026-24240 HIGH
7.8 Jul 01

Deserialization of untrusted data in NVIDIA Megatron Bridge for Linux (CWE-502) can lead to arbitrary code execution, pr

CVE-2026-24249 HIGH
7.8 Jul 01

Deserialization of untrusted data in NVIDIA Megatron Bridge for Linux allows a low-privileged local attacker to achieve

CVE-2026-24251 HIGH
7.8 Jul 01

Local code execution and privilege escalation in NVIDIA Megatron Bridge (Linux) stems from unsafe handling of dynamicall

CVE-2026-24245 HIGH
7.8 Jul 01

Arbitrary code execution in NVIDIA Megatron Bridge for Linux arises from unsafe deserialization of untrusted data (CWE-5

CVE-2026-24244 HIGH
7.8 Jul 01

Arbitrary code execution and privilege escalation in NVIDIA Megatron Bridge on Linux arises from unsafe deserialization

CVE-2026-24242 HIGH
7.8 Jul 01

Server-side request forgery in NVIDIA Megatron Bridge for Linux allows an attacker to coerce the software into issuing a

CVE-2025-33240 HIGH
7.8 Feb 18

NVIDIA Megatron Bridge contains a vulnerability in a data shuffling tutorial, where malicious input could cause a code i

CVE-2025-33239 HIGH
7.8 Feb 18

NVIDIA Megatron Bridge contains a vulnerability in a data merging tutorial, where malicious input could cause a code inj

Share

EUVD-2026-41019 vulnerability details – vuln.today

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy