CVE-2026-24228: NVIDIA NeMo Framework for Linux contains a vulnerability where an attacker may cause deserialization of untrusted data

Synopsis

Deserialization of untrusted data vulnerability in NVIDIA NeMo Framework for Linux allows a local, low-privileged attacker to supply a maliciously crafted payload that the framework deserializes without sufficient validation. The attacker needs an existing low-privilege account on the host and does not require network access or victim interaction. Successful exploitation enables arbitrary code execution, privilege escalation, data tampering, and sensitive information disclosure. HarborGuard tracks this advisory and will make a patched-image rebuild available the moment an upstream fix is published.

HarborGuard Coverage

Exploit Conditions

• Network reachabilityNot required

  The attacker needs an existing shell or process on the host; no network exposure is required to reach the vulnerable code path.

• AuthenticationRequired

  Any low-privilege local account is sufficient; the attacker does not need administrative or root credentials to trigger the deserialization.

• Victim interactionNot required

  No user action or social engineering is needed; the attacker can trigger exploitation without any participation from another user.

• Attack complexityDetail

  The exploit is reliable and condition-free; no race condition, memory-layout dependency, or other environmental factor must be satisfied.

Blast Radius

• Executes arbitrary code in the context of the NeMo Framework process, enabling full host compromise if the process runs with elevated permissions.
• Escalates privileges beyond the initial low-privilege account, potentially reaching root or container-escape level access.
• Modifies persisted model weights, training data, or application configuration stored on the host.
• Reads sensitive data accessible to the NeMo process, including credentials, API keys, and proprietary model artifacts.