Single IP Behind 83% of Ivanti EPMM Attacks, Sleepers Found
GreyNoise traces Ivanti EPMM exploitation to bulletproof hosting on PROSPERO network. Defenders find dormant webshells—signs of initial access broker activity.
A single IP address on bulletproof hosting infrastructure is responsible for the vast majority of exploitation attempts against Ivanti Endpoint Manager Mobile (EPMM), according to new telemetry from GreyNoise. The finding comes as researchers discover dormant "sleeper" webshells on compromised systems—a tactic consistent with initial access brokers preparing to sell footholds.
We covered the initial Ivanti EPMM zero-day disclosure when CISA added it to the KEV catalog in late January. The exploitation picture has since become much clearer.
Attack Attribution and Infrastructure
GreyNoise recorded 417 exploitation sessions from 8 unique source IP addresses between February 1-9. One address—193.24.123[.]42—accounted for approximately 346 of those sessions, representing 83% of all observed attempts.
The malicious IP operates on infrastructure provided by PROSPERO, an autonomous system (AS200593) assessed as linked to Proton66. This network has previously served as distribution infrastructure for GootLoader, Matanbuchus, SpyNote, and Coper malware.
What makes this particularly concerning: the same host simultaneously attacked four unrelated CVEs during the same window, including Oracle WebLogic, GNU InetUtils telnetd, and GLPI. This pattern suggests automated scanning infrastructure rather than a targeted campaign.
Sleeper Webshell Campaign
Researchers at Defused Cyber reported a troubling finding: some compromised EPMM instances contain dormant in-memory Java class loaders rather than active backdoors. These "sleeper shells" remain inactive until triggered—a hallmark of initial access broker operations.
The tactic makes sense economically. Brokers establish persistent access to high-value targets, then sell that access on dark web marketplaces to ransomware operators or nation-state actors. A dormant implant is harder to detect during routine security monitoring.
Organizations that patched the CVE-2026-1281 vulnerability should not assume they're clean. Post-exploitation artifacts may persist even after the vulnerability is closed.
Confirmed Victims
Several European government agencies have confirmed targeting:
- Netherlands' Dutch Data Protection Authority (AP)
- Council for the Judiciary (Rvdr)
- European Commission
- Finland's Valtori
The Dutch DPA and Rvdr confirmed their EPMM instances were breached on or before January 29—likely before patches were available.
Evasion Techniques
The primary attacker employed more than 300 unique user agent strings spanning Chrome, Firefox, Safari, and multiple operating system variants. This indicates automated tooling designed to evade basic fingerprinting defenses.
Notably, 85% of sessions involved DNS beaconing without payload deployment or data exfiltration. The attackers appear focused on establishing initial access and validating exploitability rather than immediate data theft.
Detection Guidance
Network defenders should:
- Review DNS logs for out-of-band callbacks to unfamiliar domains
- Search web logs for requests to
/mifs/403.jsppaths - Block AS200593 (PROSPERO) at the network perimeter
- Hunt for in-memory Java artifacts that persist across EPMM restarts
- Apply patches immediately if not already done—Ivanti notes no downtime is required
The broader pattern here fits what we've seen with other edge device vulnerabilities—internet-facing management interfaces remain prime targets for both opportunistic scanning and targeted exploitation.
Why This Matters
Mobile device management platforms like EPMM sit at a critical junction in enterprise architecture. They have visibility into managed endpoints, store credentials, and often bypass standard network segmentation. A compromised MDM server gives attackers a launching point into the devices it manages.
The sleeper webshell discovery is particularly troubling. Organizations may believe they've contained the incident after patching, only to find weeks later that access was sold to a ransomware crew. Security teams should treat EPMM compromise as an ongoing hunt rather than a closed incident.
Related Articles
Ivanti EPMM Zero-Days Under Attack, CISA Sets Friday Deadline
Two critical code injection flaws in Ivanti Endpoint Manager Mobile enable unauthenticated RCE. Federal agencies must remediate by February 1.
Jan 30, 2026Christmas Attack Campaign Targets ColdFusion with 2.5M Requests
GreyNoise researchers uncover coordinated campaign exploiting 767 CVEs across 47 technology stacks. Hong Kong-based infrastructure generated 98% of attack traffic on Christmas Day.
Jan 5, 2026Claude Code Flaws Let Malicious Repos Steal API Keys, Run Code
Check Point found CVE-2025-59536 and CVE-2026-21852 in Anthropic's Claude Code. Opening a cloned repo could execute code and leak API credentials.
Feb 26, 2026OpenLIT GitHub Actions Flaw Exposes API Keys, Cloud Credentials
CVE-2026-27941 (CVSS 9.9) lets attackers execute code via pull requests to OpenLIT, stealing GITHUB_TOKEN and cloud secrets. Patch to 1.37.1 now.
Feb 26, 2026