NVIDIA✓

Stav AI Act assessment

Editorial assessment, not legal advice. Stav's risk ratings, scores, and verdicts are our own analysis of publicly available information and may be incomplete or out of date. Verify independently before making compliance or procurement decisions.

• securityUnknown

  NVIDIA's AI Trust Center confirms ISO 27001, ISO 27017, ISO 27018, ISO 50001, SOC (SSAE 18), SIG Lite, and TISAX certifications — a comprehensive and current set of third-party validated security controls, updated as recently as June 2026.

• transparencyUnknown

  NVIDIA publishes open model weights, training data information, and model cards for the Nemotron family and other models on HuggingFace (huggingface.co/nvidia), providing meaningful transparency for open-weights models.

• transparencyUnknown

  NVIDIA publishes a comprehensive, actively maintained privacy policy with EU-localised variants (en-eu) and a dedicated terms-of-service page, both hosted on its own canonical domain.

• complianceUnknown

  NVIDIA was the first company accredited by ANAB to establish the Halos AI Systems Inspection Lab, creating a certified framework for functional safety, cybersecurity, and AI compliance for physical AI deployments.

• stabilityUnknown

  NVIDIA is a well-established, publicly listed US corporation with CEO Jensen Huang holding the role continuously since co-founding the company in 1993 — an exceptional leadership continuity record for a major technology organisation.

• communityUnknown

  NVIDIA NIM microservices are deeply integrated with HuggingFace Inference Endpoints and support over 100,000 HuggingFace LLMs for rapid enterprise deployment, demonstrating broad community and ecosystem engagement.

• complianceUnknown

  NVIDIA's AI Trust Center (nvidia.com/en-us/ai-trust-center/) is a dedicated, publicly accessible compliance and trust resource that covers security certifications, responsible AI, and physical AI safety — updated as recently as June 2026.

Published safeguards & certifications

Privacy policy

https://www.nvidia.com/en-us/about-nvidia/privacy-policy/

Flagged issues

1

• NVIDIA's GDN service-specific terms contain a broad 'Residuals' clause granting NVIDIA unrestricted use of information retained in the memory of individuals who have accessed customer confidential information. Enterprise customers should assess the data processing implications.low

Privacy policy review

Country

🇺🇸 United States

Open-weights policy

Mixed

Models in catalogue

0

Website

https://www.nvidia.com/en-us/ai/

HuggingFace

nvidia · Verified · 56,745 followers

HF published

776 models · 234 datasets · 96 papers

Creator profile

NVIDIA is a US-incorporated, publicly listed technology company headquartered in Santa Clara, California, making it fully subject to CLOUD Act and FISA Section 702 compelled-access obligations — a material consideration for EU regulated-sector deployments. Its AI model portfolio (the Nemotron family) spans genuinely open-weights releases under a permissive NVIDIA Open Model License with published training data and recipes, alongside proprietary NIM microservices requiring an enterprise licence. The 2022 Lapsus$ breach — which exposed over 70,000 employee credentials and proprietary source code — is the most significant security incident on record; NVIDIA has since invested in an AI Trust Center, SOC-audited controls, and a published product-security programme, but no ISO 27001 certification has been publicly confirmed for its AI services stack.

Stav editorial summary