CVE-2026-53805: NVIDIA SIL GEN3C Unauthenticated RCE via Pickle Deserialization in Inference API

Published Jun 17, 2026
·
Updated

NVIDIA Spatial Intelligence Lab's (SIL) GEN3C contains an unauthenticated remote code execution vulnerability in the inference API server where the /request-inference and /seed-model endpoints deserialize raw HTTP request bodies using Python's pickle.loads() without authentication or input validation. Attackers can supply a crafted payload containing a reduce gadget to the inference API port to achieve remote code execution as the inference process.

Affected Software

1 affected component
Nvidia SIL GEN3C

Remediation

Recommended actions to resolve this vulnerability, in priority order.

  1. Configuration

    Stop deserializing raw HTTP request bodies with pickle.loads; do not accept Python pickle payloads from untrusted sources and replace unsafe unpickling with a safe serialization/deserialization approach or only deserialize trusted, authenticated input.

    GEN3C inference API (/request-inference, /seed-model) use_pickle_deserialization = false
  2. Configuration

    Require authentication and authorization for the /request-inference and /seed-model endpoints to prevent unauthenticated access.

    GEN3C inference API (/request-inference, /seed-model) authentication_required = true
  3. Configuration

    Validate and sanitize request bodies before any deserialization; explicitly reject or block serialized Python pickle payloads and other unsafe input formats from untrusted clients.

    GEN3C inference API (/request-inference, /seed-model) input_validation = enabled
  4. Compensating control

    Restrict network access to the inference API port using firewall rules, ACLs, WAF, or network segmentation so only trusted IPs or management networks can reach the endpoints.

  5. Operational

    If the inference API was exposed, investigate for signs of compromise (containment, forensic analysis), and remediate any confirmed compromise (restore services from known-good backups, rebuild affected hosts).

Event History

Jun 17, 2026
CVE Published
via MITRE·04:44 PM
Data Sourced
via MITRE·04:44 PM
DescriptionSeverityWeakness
Data Sourced
via NVD·06:18 PM
DescriptionSeverityWeakness
Jul 13, 2026
Exploit Published
via GitHub·07:29 AM

Frequently Asked Questions

1

What is the severity of CVE-2026-53805?

CVE-2026-53805 has a critical severity score of 9.8.

2

What is CVE-2026-53805?

CVE-2026-53805 is a remote code execution vulnerability in NVIDIA SIL GEN3C's inference API due to unauthenticated deserialization of HTTP request bodies.

3

How do I fix CVE-2026-53805?

To fix CVE-2026-53805, ensure proper authentication and input validation for the /request-inference and /seed-model endpoints.

4

What can attackers do with CVE-2026-53805?

Attackers can exploit CVE-2026-53805 to execute arbitrary code on the server by sending crafted requests to the affected endpoints.

5

Is CVE-2026-53805 exploitable remotely?

Yes, CVE-2026-53805 is an unauthenticated remote code execution vulnerability, making it exploitable from any remote location.

Contact

SecAlerts Pty Ltd.
132 Wickham Terrace
Fortitude Valley,
QLD 4006, Australia
info@secalerts.co
By using SecAlerts services, you agree to our services end-user license agreement. This website is safeguarded by reCAPTCHA and governed by the Google Privacy Policy and Terms of Service. All names, logos, and brands of products are owned by their respective owners, and any usage of these names, logos, and brands for identification purposes only does not imply endorsement. If you possess any content that requires removal, please get in touch with us.
© 2026 SecAlerts Pty Ltd.
ABN: 70 645 966 203, ACN: 645 966 203