Skip to content

English

Places to contact

NVIDIA

Type: AI company
Place: United States — USA
Last checked: 2026-09-22
Next check due: 2027-03-21

Ways to reach them

NVIDIA PSIRT · Email address
psirt@nvidia.com
Open contact route · 2026-09-22
Who may use it: public
Evidence that this route accepts contact · 2026-09-22
Product security vulnerabilities and AI concerns.
PGP key at https://www.nvidia.com/en-us/security/pgp-key/
Report a Security Vulnerability or NVIDIA AI Concern · Responsible disclosure programme
https://app.intigriti.com/company/programs/nvidia/nvidiavdp/detail
Open contact route · 2026-09-22
Who may use it: public
Evidence that this route accepts contact · 2026-09-22
Explicitly includes 'AI vulnerability or hack, AI trust concern, malicious prompt engineering, data poisoning possibility'.
One of the very few intake forms in this list that names AI trust concerns as a first-class category rather than excluding them.
NVIDIA public bug bounty (Intigriti) · Bug or safety bounty
https://app.intigriti.com/company/programs/nvidia/nvidiapublicbugbounty/detail
Open contact route · 2026-09-22
Who may use it: public
Evidence that this route accepts contact · 2026-09-22
Paid security research.
Homepage · Homepage
https://www.nvidia.com/en-us/security/
Verification expired
Evidence that this route accepts contact
NVIDIA Frontier AI Risk Assessment (17 February 2025) · Published policy or framework
https://images.nvidia.com/content/pdf/NVIDIA-Frontier-AI-Risk-Assessment.pdf
Verification expired
Evidence that this route accepts contact

What it does

NVIDIA PSIRT handles product security including AI-specific concerns; a Trustworthy AI function publishes model cards and the NeMo Guardrails tooling.

Honest assessment

No public complaints found about NVIDIA PSIRT being unresponsive; PSIRT is an established, publicly active team (it posts as @nvidiapsirt). The AI-concern category on the VDP intake form is genuine and unusual. Note that from 1 October 2026 NVIDIA PSIRT will publish security bulletins only on GitHub (github.com/nvidia/product-security), so that is where to watch for outcomes.

Notes

The intake form's 'NVIDIA Artificial Intelligence (AI) Concern' option is the route to use — it is a real dropdown, not marketing. No explicit safe-harbour language was visible on the report-vulnerability page; check the PSIRT policies page before doing anything aggressive.

Sources

https://app.intigriti.com/company/programs/nvidia/nvidiavdp/detail · 2026-09-22
https://app.intigriti.com/company/programs/nvidia/nvidiapublicbugbounty/detail · 2026-09-22

Something wrong here?