No Patch for New PhantomRPC Privilege Escalation Technique in Windows

Summary

Kaspersky researcher Haidar Kabibo discovered PhantomRPC, an architectural weakness in Windows Remote Procedure Call (RPC) that allows local privilege escalation to System level. The vulnerability exploits the fact that Windows does not verify RPC server legitimacy, enabling attackers who compromise a privileged service to deploy a fake RPC server and impersonate legitimate services. Microsoft has classified the issue as moderate-severity and stated it does not require immediate remediation, though the weakness potentially affects all Windows versions and creates a large attack surface through multiple exploitation paths.

Full text

A vulnerability in the Windows Remote Procedure Call (RPC) mechanism allows attackers to elevate their privileges to System, Kaspersky reports. The local privilege escalation issue potentially affects all Windows versions and abuses another legitimate Windows mechanism, where processes are allowed to impersonate other processes to perform specific actions. The root cause of the security defect, which Kaspersky researcher Haidar Kabibo named PhantomRPC, is an architectural weakness, potentially turning any process that depends on RPC into a possible escalation path. In Windows, RPC is the mechanism that allows processes to communicate with one another and invoke functions that are implemented in other processes, regardless of their execution contexts. It uses a client–server model, where the invoking process is the client. Windows also allows services to impersonate users or other services to temporarily operate in their security context, and controls this functionality through impersonation levels ranging from Anonymous to Impersonate and Delegate. To impersonate a client, the service needs a specific privilege that is granted by default to certain services, such as those running under the Local Service and Network Service accounts.Advertisement. Scroll to continue reading. Furthermore, the RPC runtime does not verify the legitimacy of RPC servers, and processes are allowed to deploy RPC servers exposing the same endpoints as legitimate services. To exploit PhantomRPC, Kabibo says, an attacker needs to compromise a privileged service, deploy a fake RPC server, listen to specific requests, and then impersonate the targeted service to escalate their privileges. Network Service account service abuse The attacker could compromise a service running under the Network Service account and deploy a fake RPC server with the RPC interface UUID and exposed endpoint name as TermService, the default Remote Desktop service. The attacker could then force a policy update to cause the Group Policy service, which runs with System privileges, to perform an RPC call to TermService. Because TermService is disabled by default, the request would fail. However, the attacker’s RPC server, which also receives the RPC request, can now impersonate the security context of the Group Policy service and elevate privileges to System. After identifying other RPC clients attempting to communicate with unavailable servers, Kabibo discovered four other PhantomRPC exploitation paths, noting that the weakness leads to a large attack surface, because numerous system DLLs in Windows rely on RPC. “Applications that invoke seemingly benign APIs may unintentionally trigger privileged RPC interactions. Under certain conditions, these interactions could be abused to achieve local privilege escalation without the user’s knowledge,” the researcher says. In another scenario, the attacker’s fake RPC server would wait for a high-privileged user to launch Microsoft Edge, which makes an RPC call to the TermService upon start. The attacker’s server intercepts the request and elevates its privileges from Network Service to System. Another attack path listens to the background RPC calls that the Diagnostic System Host Service (WDI) periodically makes to TermService using a high impersonation level. Using the same setup, the attacker elevates privileges without user interaction, as the WDI automatically makes the calls every 5 to 15 minutes. Local Service account service abuse The security researcher also discovered two attack paths that abuse a Local Service account to escalate privileges, such as the DHCP Client service, which is enabled by default and exposes an RPC server with multiple interfaces and endpoints. The attacker’s fake RPC server mimics the legitimate RPC service exposed by the DHCP Client and listens for the RPC calls that ipconfig makes to it when run by an administrator. The scenario assumes that the DHCP Client service is disabled, allowing the fake server to impersonate the client. The Windows Time service, also enabled by default under the Local Service account, exposes an RPC server with two endpoints, and the executable w32tm.exe interacts with it using RPC. Because w32tm.exe calls a nonexistent named pipe not exposed by the legitimate service, the attacker can deploy an RPC server that exposes it, then wait for a high-privileged user to run the executable so that the RPC request is redirected to the malicious server. “In this scenario, it is important to note that the legitimate Windows Time service does not need to be disabled. Because the executable attempts to connect to a nonexistent endpoint, it is sufficient for the attacker to expose that endpoint through the malicious RPC server,” the researcher says. Kaspersky reported the issue in September 2025. Microsoft classified it as moderate-severity due to the required impersonation privilege and said it does not require immediate remediation. SecurityWeek has emailed Microsoft for a statement and will update this article if the company responds. Related: Incomplete Windows Patch Opens Door to Zero-Click Attacks Related: Organizations Warned of Exploited Windows, Adobe Acrobat Vulnerabilities Related: In Other News: Cyberattack Stings Stryker, Windows Zero-Day, China Supercomputer Hack Related: ClickFix Attack Uses Windows Terminal to Evade Detection Written By Ionut Arghire Ionut Arghire is an international correspondent for SecurityWeek. More from Ionut Arghire UNC6692 Uses Email Bombing, Social Engineering to Deploy ‘Snow’ MalwareEasily Exploitable ‘Pack2TheRoot’ Linux Vulnerability Leads to Root AccessChina-Linked APT GopherWhisper Abuses Legitimate Services in Government AttacksPre-Stuxnet Sabotage Malware ‘Fast16’ Linked to US-Iran Cyber TensionsUS Federal Agency’s Cisco Firewall Infected With ‘Firestarter’ BackdoorBitwarden NPM Package Hit in Supply Chain AttackCloudsmith Raises $72 Million in Series C FundingRilian Raises $17.5 Million for AI-Native Security Orchestration Latest News Electric Motorcycles and Scooters Face Hacking Risks to Security and Rider SafetyGermany Suspects Russia Is Behind Signal Phishing That Targeted Top OfficialsSpectrum Security Emerges From Stealth Mode With $19 MillionMedtronic Hack Confirmed After ShinyHunters Threatens Data LeakIncomplete Windows Patch Opens Door to Zero-Click AttacksOpenSSH Flaw Allowing Full Root Shell Access Lurked for 15 YearsMalicious AI Prompt Injection Attacks Increasing, but Sophistication Still Low: GoogleEnergy and Water Management Firm Itron Hacked Trending Daily Briefing Newsletter Subscribe to the SecurityWeek Email Briefing to stay informed on the latest threats, trends, and technology, along with insightful columns from industry experts. Webinar: A Step-by-Step Approach to AI Governance April 28, 2026 With "Shadow AI" usage becoming prevalent in organizations, learn how to balance the need for rapid experimentation with the rigorous controls required for enterprise-grade deployment. Register Virtual Event: Threat Detection and Incident Response Summit May 20, 2026 Delve into big-picture strategies to reduce attack surfaces, improve patch management, conduct post-incident forensics, and tools and tricks needed in a modern organization. Register People on the MoveNeill Feather has been named Chief Executive Officer at Point Wild.Oasis Security has appointed Michael DeCesare as President.Sterling Wilson has joined IGEL as Global Field CTO, Business Continuity and Disaster Recovery.More People On The MoveExpert Insights Why Cybersecurity Must Rethink Defense in the Age of Autonomous Agents From autonomous code generation to decision-making systems that initiate actions without human intervention, the industry is entering a new phase. (Torsten George) Government Can’t Win the Cyber War Without the Private Sector Securing national resilience now depends on faster, deeper partnerships with the private sector. (Steve Durbin) The Hidden ROI of Visibility: Better Decis

Indicators of Compromise

• mitre_attack — T1134 - Access Token Manipulation
• mitre_attack — T1548.004 - Abuse Elevation Control Mechanism: Elevated Execution with Prompt

Entities