Nvidia provides GPUs, drivers, and software used in AI and computing; flaws in these components can expose systems, data, and workloads.
Search across headline titles and summaries.
Background for this topic.
Nvidia develops graphics processing units (GPUs), accelerator cards, system-on-chip platforms, and the drivers and software stacks that control them. Its hardware is used in workstations, cloud systems, high-performance computing, and AI infrastructure; security news under this tag therefore commonly concerns device firmware, kernel drivers, GPU runtimes, management tools, and software libraries rather than the silicon alone.
Security advisories matter because flaws in drivers or privileged GPU components can allow local code to crash systems, gain elevated access, or cross intended isolation boundaries, depending on the affected platform. Shared GPU servers also require careful tenant and data isolation: residual data in device memory or insecure accelerator-management interfaces can expose workloads. Operators should track Nvidia security bulletins, inventory driver and firmware versions, obtain updates through trusted channels, restrict management endpoints, and test upgrades against dependent CUDA or AI workloads. Vulnerability assessment should include container and orchestration integrations, since a GPU-enabled workload may receive additional host access.
Weekly headline count for the current query.
Attacks by artificial intelligence agents are a reality. Experts at Nvidia's GTC conference say defenders need to use the same tools to fight them off.
In partnership with Emirates tech company G42, Microsoft is building the first stage of a 5-gigawatt US-UAE AI campus using Nvidia GPUs.
Security researchers discovered multiple vulnerabilities in AI infrastructure products, including one capable of remote code execution.
The US banned the sale of AI chips to China and then backed off. Now, Chinese sources are calling on NVIDIA to prove its AI chips have no backdoors.
The flaws in the company's Triton Inference Server enables model theft, data leaks, and response manipulation.
A container escape flaw involving the NVIDIA Container Toolkit could have enabled a threat actor to access AI datasets across tenants.
Nvidia's DOCA Argus prevents attacks before they compromise AI architectures.
A fix for a critical flaw in a tool allowing organizations to run GPU-accelerated containers released last year did not fully mitigate the issue, spurring the need to patch a secondary flaw to protect organizations that rely on NVIDIA processors for AI workloads.
Nvidia doesn't just make the chips that accelerate a lot of AI applications — the company regularly creates and uses its own large language models, too.
Nvidia's latest GPUs are a hot commodity for AI, but security vulnerabilities could expose GPUs — which could be up to seven years old — to attacks from hackers.
With companies pouring billions into AI software and hardware, these installations need to be protected from cybersecurity threats and other security lapses.
Nvidia and AMD do face expanded export rules for their A100 and H100 artificial intelligence (AI) chips in the Middle East, but it's not yet clear why.
If unpatched, a host of GPU Display Driver flaws could expose gamers, graphic designers, and others to code execution, denial of service, data tampering, and more.
The new open source specification from Open Compute Project is backed by Google, Nvidia, Microsoft, and AMD.