Vercel Verifies Critical KVM Zero‑Day Following Researcher’s Host‑Root Escape Demo

Vercel confirmed a critical zero‑day vulnerability in the Linux KVM hypervisor that allows a guest virtual machine to escape and obtain root access on the host.

Independent security researcher Paulos Yibelo first disclosed the flaw, providing proof‑of‑concept code that demonstrates a complete VM escape. Yibelo explained that the exploit lets code executing inside a guest seize administrative control of the host.

KVM (Kernel‑based Virtual Machine) underpins a significant share of cloud and serverless workloads, Vercel’s platform among them. Such an escape undermines the isolation assurances essential to multi‑tenant setups, possibly letting an attacker view or modify other customers’ data or compromise the host infrastructure.

After the report, Vercel’s security team performed its own analysis, verified the problem, and worked with the upstream KVM maintainers to create a patch. The firm granted Yibelo a $50,000 bounty via its vulnerability‑responsibility program.

Vercel announced that patches are already rolled out across its production services and urged customers to update any self‑managed KVM‑based deployments. The company also noted that, to date, there is no indication the flaw was exploited in the wild prior to disclosure.

The episode underscores the ongoing difficulty of protecting virtualization layers as workloads increasingly move to container‑ and function‑as‑a‑service models. A growing attack surface is prompting cloud providers and open‑source projects to speed up security audits and patch cycles.

Experts point out that coordinated disclosure and swift remediation, as shown here, are crucial for reducing exposure time. Vercel said it will continue collaborating closely with the KVM community to watch for similar issues and to bolster its own hardening measures.