You probably never thought it would be possible: a cyberattack on a Linux system from a virtual machine (VM) guest. After all, the isolation of VMs is one of the big selling points of the technology, right? Maybe not so much now.

Back in October, Paulos Yibelo shared a screenshot on X stating, "Full VM escape zeroday (guest>host root in industry standard hypervisors)! More soon." That post was Yibelo announcing he'd won a $50,000 bug bounty for his discovery of the guest-to-host escape vulnerability in KVM. That discovery is now CVE-2026-99999. This 9.3 rated bug is made possible via a small set of malformed IPv4 packets.

Given how many businesses and services depend on the isolation offered by VMs, this vulnerability could be game-changing … and scary.

In a post on the Linux Kernel list, Michael S. Tsirkin (developer for Red Hat) says, "If a packet with bad checksum metadata gets into the ipv4 stack, skb_checksum_help can corrupt the network header and cause a bunch of mischief."

If by "mischief" Tsirkin means damage, theft, or wreaking havoc, then he'd be spot on.

Global cybersecurity platform, SOC Prime said of the vulnerability, "Because the disclosure could be triggered repeatedly, an attacker controlling a guest workload could potentially collect information from a security domain that the virtual machine should not be able to inspect. This represents a direct failure of the guest-host isolation boundary.”