CVE-2025-23290 is a vulnerability identified in NVIDIA's Virtual GPU (vGPU) software, specifically within the Virtual GPU Manager component of the GPU display drivers versions R535 and R570. The flaw is categorized under CWE-200, indicating an exposure of sensitive information to unauthorized actors. The vulnerability arises because a guest virtual machine (VM) can access global GPU metrics that are influenced by workloads running in other VMs on the same physical host. This cross-VM information leakage could allow an attacker controlling a guest VM to infer or extract sensitive data about the operations or workloads of other co-resident VMs sharing the same GPU resources. The vulnerability does not allow modification of data or disruption of service but compromises confidentiality by exposing potentially sensitive performance or usage metrics. The CVSS v3.1 base score is 2.5, reflecting a low severity primarily due to the requirement for local access (AV:L), high attack complexity (AC:H), and the need for low privileges (PR:L) but no user interaction (UI:N). There are no known exploits in the wild, and no patches have been linked yet, indicating this is a newly disclosed issue. The vulnerability is limited to specific NVIDIA GPU driver versions used in virtualized environments leveraging vGPU technology, which is common in data centers and cloud infrastructures that provide GPU acceleration to multiple tenants or workloads via virtualization.

For European organizations, the impact of CVE-2025-23290 is primarily related to confidentiality risks in multi-tenant virtualized environments that utilize NVIDIA vGPU technology. Organizations running virtualized workloads on NVIDIA GPUs, such as cloud service providers, research institutions, financial firms, and enterprises with GPU-accelerated virtual desktops or compute clusters, could face information leakage between VMs. This could lead to exposure of sensitive operational metrics or workload characteristics, potentially aiding attackers in profiling or planning further attacks. However, the vulnerability does not allow direct data modification or denial of service, limiting its impact to information disclosure. Given the low CVSS score and the complexity of exploitation, the immediate risk is moderate to low, but it is more significant in environments with strict confidentiality requirements or where sensitive workloads are co-located on shared GPU infrastructure. European organizations with compliance obligations around data privacy and confidentiality (e.g., GDPR) should consider the implications of any cross-tenant data leakage, even if indirect.

To mitigate CVE-2025-23290, European organizations should: 1) Monitor NVIDIA's official channels for patches or updates addressing this vulnerability and apply them promptly once available. 2) Restrict access to virtualized GPU resources by enforcing strict tenant isolation policies and limiting the number of co-resident VMs sharing the same physical GPU, reducing the attack surface. 3) Employ network segmentation and access controls to limit which users or systems can deploy or manage vGPU-enabled VMs. 4) Audit and monitor GPU usage metrics and logs for unusual access patterns that could indicate exploitation attempts. 5) Consider disabling vGPU sharing features temporarily in highly sensitive environments until a patch is available. 6) Engage with NVIDIA support or vendors to understand any recommended configuration changes or mitigations specific to affected driver versions. 7) Incorporate this vulnerability into risk assessments and incident response plans, especially for cloud or virtualized GPU deployments.