NVIDIA Brev cloud platform breached, SSH keys potentially exposed

The breach was disclosed in a notification sent to affected customers, with NVIDIA stating that it had investigated the incident and identified unidentified activity on the platform. “We are writing to inform you that NVIDIA Corporation has experienced a security incident that may have affected the security of your data,” the notification reads. “Our analysis determined that your information from our Brev network may have been affected. This SSH key has since been rotated, and your instances have been updated with the replacement key.”

SSH is a protocol used to establish a secure connection between virtual machines or remote computers, and developers can use it to manage servers and transfer files without physical access. Brev uses SSH keys to allow customers to access cloud-based development services; if attackers obtained these keys they could potentially take control of a virtual machine, depending on permissions and other controls.

NVIDIA did not explain how the attackers gained access or whether any customer data or workloads were accessed. The company said it investigated the incident and found no evidence that the exposed keys had been misused.

The incident comes as several firms hosting AI development services and cloud infrastructure grapple with mounting security incidents involving customer credentials. Cybersecurity firm Wiz published a blog on October 30 dubbing the month “Black October” for AI developer platforms, identifying at least nine platforms that suffered security incidents, including Zeabur, Fireworks AI, Lightning AI, Opencode, Warp and Northflank. Wiz researchers noted that “all public reports highlight targeted exploitation, explicitly hunting stored environment variables, with a focus on those granted AI access.” The firm shared two cases of compromised Amazon Bedrock credentials where attackers generated roughly 800,000 API calls over 30 hours in one instance and about 1.7 million calls over 13 hours in another, with combined usage reaching an estimated $100,000 at list prices.

Such incidents are not just restricted to NVIDIA, with major US banks including JP Morgan, Morgan Stanley and Bank of America restricting AI agents from accessing customer accounts this month. Their primary concern is that automated assistants could veer off set instructions and initiate unauthorised approvals. Separately, the CyberCartel hacking group has used Anthropic’s Claude and xAI’s Grok to attack more than 400 Latin American institutions, mostly banks and mobile payment services, after obtaining customer login credentials. China-linked attackers have also been accused of using AI agents in a ConnectWise supply-chain attack that exposed about 2,250 Microsoft 365 tenants, with some attacks reportedly reaching privileged access in under three hours.

Bradford Regeski, a threat intelligence analyst at RH-ISAC, recommended extensive checks for customers in SSH-related incidents, asking users to “Isolate hosts that installed the compromised extension during the exposure window,” remove unauthorized SSH keys and persistence mechanisms, and rotate credentials and sessions.