Summary:
A ransomware attack linked to BlackNevas destroyed an organization's VMware environment during a 10-day intrusion, affecting four ESXi hosts, every hosted virtual machine, two vCenter Server Appliances, and connected backup infrastructure. Securelist's investigation, published on November 13, 2026, found no exploited virtualization vulnerability: the attackers used credentials stolen from systems outside the virtualization environment to reach vCenter and then abused legitimate management features, including vCenter Host Profiles, to enable SSH on ESXi hosts and deploy the payload. Before the final stage, they disabled backup antivirus, stopped backup services, and encrypted recovery infrastructure, leaving defenders without a restoration path. Attribution to BlackNevas rests on the ransom notes, as the ESXi payload was unavailable for analysis.
Technical Description:
Securelist researchers Elsayed Elrefaei and Mohamed Ghobashy investigated an attack in which the BlackNevas name appeared in the ransom notes. The investigators found no guest-to-host escape or hypervisor memory corruption; instead, credentials stolen from systems outside the virtualization environment opened access to vCenter, and the attackers used valid accounts and legitimate management interfaces throughout, which made their activity harder to distinguish from normal administration. The ESXi payload was not available for analysis, which limits binary-level attribution and verification of some command-line options.
The attackers first gained administrative execution on backup infrastructure and extracted stored credentials, turning those systems into a bridge into other infrastructure. On the disaster-recovery vCenter appliance they authenticated as root through SSH and deployed the MeshAgent remote-management tool just 43 seconds later, then returned about four hours later through MeshAgent and installed a persistent Ligolo-ng reverse tunnel using a VMware-looking systemd service. The details and technicalities of the attack campaign are discussed further,
Delivery and Infection Chain:
The intrusion relied on valid administrative credentials rather than vulnerability exploitation, so each step resembled normal management activity. Security agents inside virtual machines cannot directly observe the hypervisor process modifying their virtual disks, which created a monitoring gap. The infection chain was identified as follows,
- Credentials stolen from systems outside the virtualization environment, and others extracted from backup infrastructure after administrative execution, give the attackers access to vCenter and backup systems.
- On the disaster-recovery vCenter appliance, the attackers log in as root over SSH, deploy MeshAgent, and later install a persistent Ligolo-ng reverse tunnel as a service resembling VMware software.
- The attackers abuse vCenter Host Profiles to change the TSM-SSH service state, enabling direct SSH on ESXi hosts in about eleven minutes, with later connections originating from vCenter itself.
- Antivirus in the backup environment is disabled days earlier, then backup services are stopped and recovery infrastructure is encrypted, removing the expected restoration path.
- The ESXi payload arrives over SFTP, runs a roughly 16-second self-test about three hours before execution, and then wipes VMFS storage, overwriting the first 2 KB of virtual-disk files and stopping all VMs on the disaster-recovery host within 23 seconds.
Technical Capabilities:
The attackers abused vCenter Host Profiles, a legitimate feature for enforcing host configurations, to enable SSH on ESXi hosts, turning a compliance mechanism into an access path to the hypervisor. The payload ran with wiping and directory-traversal options against VMFS storage, and artifacts indicated that roughly the first 2 KB of virtual-disk files had been overwritten. Such damage can prevent virtual machines from booting, but it does not prove every disk is unrecoverable, as surviving backup partition metadata and other damage determine recovery prospects.
On the disaster-recovery host, all virtual machines stopped within 23 seconds, although investigators could not establish the exact shutdown technique. The payload was transferred over SFTP, and authentication logs recorded a FlowSsh client identifier. Before the final stage, the attackers weakened recovery by disabling antivirus protection in the backup environment, stopping backup services, encrypting recovery infrastructure, and disrupting administrative access. The vmware-vdma service on the vCenter appliance and a Windows remote-management agent installed as the service vspbind provided persistence.
Attribution and Evolution:
BlackNevas was identified from the ransom notes, and Securelist notes that the unavailable ESXi payload limits binary-level attribution. Earlier reporting documented BlackNevas encryption tactics, while this case shows how stolen management access enabled broader destruction of virtualized infrastructure and backups. Securelist did not connect this incident to the GhostFrame ESXi credential-theft toolkit or to the exploitation of Veeam vulnerabilities.
Active Campaign and Geographic Spread:
The intrusion lasted 10 days and affected four ESXi hosts, all hosted virtual machines, two vCenter Server Appliances, and connected backup infrastructure. Initial portal access came from three IP addresses, and the remote-management and Ligolo-ng command servers used separate infrastructure. The source does not identify the victim's sector or location.
Conclusion:
The incident shows that stolen administrative credentials and trusted management access can undermine hypervisor security without any vulnerability being exploited. Organizations should separate administrative identities for production, virtualization, and backups, restrict management access, monitor vCenter Host Profile changes and SSH enablement, and test recovery procedures, as immutable storage alone is insufficient when attackers can compromise the accounts, servers, or network paths needed to restore it.
Impact:
Successful attacks result in the destruction of virtual machines, loss of backups and recovery capability, and extended outages of services running on virtualized infrastructure. Organizations face increased risks of data loss, prolonged operational disruption, and ransom demands, while the loss of recovery infrastructure can leave them without a restoration path.
IOC and Context Details:
| Topics | Details |
| Tactic Name | Initial Access, Execution, Persistence, Credential Access, Lateral Movement, Defense Evasion, Command and Control, Impact |
| Technique Name | Valid Accounts; Credentials from Password Stores; Remote Services; Remote Access Tools; Protocol Tunneling; Create or Modify System Process; Impair Defenses; Inhibit System Recovery; Data Destruction |
| Sub Technique Name | SSH; Remote Desktop Software; Systemd Service; Disable or Modify Tools |
| Attack Type | Malware |
| Targeted Applications | VMware ESXi; VMware vCenter Server Appliance (vCenter Host Profiles); backup infrastructure; VMFS storage |
| Region Impacted | Global |
| Industry Impacted | Not specified; organizations running VMware virtualized environments |
| IOC’s | **SHA-1 Hash:** 28f8da19ddcdf26e426875c7a834b34a79bc1c33; **SHA-256 Hash:** 15dc419b7fbeb50c50820a7cd27997541deafa4fa710f074bb657f0149574baf; 5a9b7a0d3f25757aea957713d7e84cabec87597b3474368328a8fd5ed7a12f78; **IPs:** 45.149.154[.]135; 91.92.242[.]121; 31.41.33[.]212; 188.240.57[.]94; **Domains:** api[.]cloudvflare[.]com; **Services:** vmware-vdma (vCenter appliance persistence); vspbind (Windows remote-management agent); **Files and Artifacts:** svhost.exe / lock.exe; MeshAgent; Ligolo-ng; FlowSsh client identifier in authentication logs |
| CVE | NA |
Recommended Actions:
- Use separate administrative identities for production, virtualization, and backup environments, and avoid shared credentials across them.
- Restrict management access to vCenter, ESXi, and backup systems to trusted networks and hardened administrative hosts, enforce multi-factor authentication, and limit SSH access on ESXi hosts.
- Monitor vCenter Host Profile changes and correlate them with TSM-SSH service enablement, new SSH connections to ESXi hosts originating from vCenter, executable staging, and mass virtual machine shutdowns.
- Monitor vCenter appliances and backup servers for unauthorized remote-management tools and tunnels, such as MeshAgent, Ligolo-ng, and the vmware-vdma and vspbind services, and block the listed IPs and domain.
- Protect and monitor backup infrastructure by alerting on disabled antivirus and stopped backup services, safeguarding stored credentials, and isolating the backup network.
- Forward authentication, shell, host-management, and vCenter task logs outside the affected environment so that attackers or a reboot cannot erase the investigation trail.
- Hunt for the listed hashes, services, and artifacts, and alert on SFTP transfers of executables to ESXi hosts and on unusual client identifiers in SSH authentication logs.
- Maintain and regularly test recovery procedures that do not depend on the same accounts, servers, or network paths as production; if compromise is suspected, isolate the management plane, rotate credentials, and preserve logs.
Reference:
