Friday, November 20, 2026

Lateral movement concepts: authentication protocols attackers abuse and how to detect them

Lateral movement concepts: authentication protocols attackers abuse and how to detect them

An educational overview of how SMB, Kerberos, WinRM, and WMI authentication flows are abused for movement inside compromised estates, written for defenders running authorized detection programs.

TopicLateral movement — attacker technique of pivoting between hosts after initial compromise
CategoryDetection engineering and adversary tradecraft analysis
Primary UseTraining blue teams to recognize abuse of NTLM, Kerberos, WinRM, and WMI in authorized lab environments
Safe UseEducational and defensive analysis for authorized security teams; reproduce concepts only in isolated labs you own
Telemetry NoteConcepts map to Windows Event Log channels such as Security 4624/4648, Sysmon Event ID 3 and 10, and PowerShell Operational logs

Lateral movement is the phase of an intrusion that sits between initial foothold and objective, and it is where most adversary tradecraft becomes visible to a defender who knows what to watch. Unlike initial access, which often happens at the perimeter through phishing or an exposed service, lateral movement happens inside the network, over protocols that the environment legitimately needs to function: SMB, Kerberos, NTLM, WinRM, WMI, and RDP. This creates the central detection problem of the phase: the traffic that carries an attacker is nearly identical to the traffic that carries a sysadmin. This essay walks through the authentication protocols most commonly abused during lateral movement, explains conceptually why each is attractive to adversaries, and outlines the detection angles defenders should instrument in authorized monitoring environments.

The foundation of Windows lateral movement is authentication, and in an Active Directory forest that means understanding the interplay between NTLM and Kerberos. NTLM is a challenge-response protocol: the client proves knowledge of a hash without ever sending it, and the domain controller brokers validation. Kerberos, by contrast, issues short-lived tickets from a Key Distribution Center after an initial pre-authentication exchange. Both protocols are legitimate, both are heavily logged, and both are abused in different ways. The defender's job is not to treat either protocol as inherently malicious but to understand the specific patterns of their use that deviate from administrative baseline — an unusual source host, a service that never authenticated before, a ticket type that does not match the account's normal behavior.

Kerberos abuse deserves particular attention because of its ticketing architecture. The classic concept here is the misuse of Ticket Granting Service tickets: once an adversary controls a service account, its long-term key can be used to forge tickets that grant persistent access, a class of technique the industry broadly refers to as golden and silver ticket concepts. From a purely defensive standpoint, the important observation is that these forgeries are invisible to the KDC itself — the domain controller never sees the forged ticket minted, only its later use when presented to a service. That means detection has to happen at the service endpoint and in Kerberos-related event logs, looking for anomalies like tickets with unusually long lifetimes, encryption types that do not match domain policy, or service accounts suddenly authenticating from new hosts.

NTLM presents a different abuse surface. Because it is a challenge-response protocol, adversaries who capture NTLM authentications in a position to relay them — classically through malicious SMB or HTTP endpoints — can forward a victim's authentication to a target service without ever knowing the password. The conceptual detection story for NTLM relay is well understood: it drives the case for disabling NTLMv1 entirely, enforcing server-side SMB signing, and monitoring for authentication events where the logon type or the negotiating host does not fit historical patterns. Defenders running authorized labs can observe this cleanly by watching Windows Event Log event 4624 with logon type 3 and comparing source workstations against a learned baseline of who authenticates where.

Remote execution channels matter as much as authentication. WinRM, Microsoft's implementation of WS-Management over HTTP/HTTPS (typically ports 5985 and 5986), has become the default remote-management path in modern Windows environments, which also makes it a favored movement channel. WMI offers a second, older channel: a rich management API that can execute code remotely and is deeply embedded in normal administrative tooling. Both generate distinctive telemetry — PowerShell Operational logs, WMI-Activity traces, and Sysmon network connections — but both are also used constantly by legitimate management agents, which is why correlation with identity context is essential rather than alerting on the protocol alone.

RDP remains the most visually obvious movement path and, correspondingly, one of the best logged. Interactive logon events, session creation, and clipboard drive-redirection settings all leave records, and abnormal RDP — a user who never touches a server suddenly logging into it at 3 a.m. — is one of the highest-signal detections available to a blue team. The broader lesson generalizes: the strength of lateral movement detection is rarely in any single protocol's logs but in joining identity, host, and time. An account authenticating over Kerberos to a new host, then opening a WinRM session, then spawning powershell.exe with an encoded command line is a chain; each link is benign alone, the sequence is not.

This chain-oriented thinking maps directly onto how mature detection programs structure their content. Frameworks like MITRE ATT&CK place lateral movement under techniques such as T1021 (remote services) with sub-techniques for SMB, WinRM, RDP, and WMI abuse, and detection engineering against those techniques is fundamentally about enumerating the telemetry sources per sub-technique. Sysmon Event ID 3 for network connections, Event ID 1 for process creation with command lines, Windows Security events 4624, 4648 (explicit credential logon), and 4672 (special privileges) form the backbone. The analytical work is deciding which combinations, given your environment's baseline, justify an alert — and accepting that volume-heavy signals like type 3 logons only become useful once they are baselined.

Credential material itself is the other half of the story. Lateral movement is only possible because credentials are reusable across hosts; techniques that harvest LSASS memory contents, extract tickets, or pull hashes from the NTDS.dit database all exist to feed the authentication flows described above. Defensively, this reframes the problem: reducing credential sprawl through tiered administration, limiting service accounts, and adopting stronger Kerberos policies such as protected users shrinks the abuse surface of every protocol simultaneously. Detection and hardening are complementary, and the most effective programs pair suspicious-authentication analytics with structural changes that make abuse harder to perform quietly.

For defenders building or tuning a program, the practical starting point is inventory: which protocols are actually in use in your estate, which accounts authenticate cross-host, and what does normal look like. From there, prioritize detections with the highest signal-to-noise: Kerberos anomalies involving service accounts, explicit credential logons (4648) followed by remote service creation, first-seen host-to-host authentication pairs, and WinRM sessions from unexpected sources. All of this work should be developed and validated in an authorized lab — a domain you own, with instrumented clients generating baseline traffic — before it is tuned against production telemetry. The concepts above are educational analysis for defensive professionals, and the goal is comprehension of adversary tradecraft so that the corresponding telemetry can be read with judgment.

The final takeaway is that lateral movement is where adversaries spend the most time and leave the most evidence, precisely because authentication must repeat at every hop. Every pivot is a new logon event, a new network connection, a new process lineage somewhere. Attackers rely on defenders not correlating those artifacts across hosts and time. Understanding the protocols — what NTLM handshakes look like, how Kerberos tickets flow, when WinRM and WMI sessions start — turns that noise into a narrative, and a defender who can read the narrative can interrupt the intrusion long before the objective is reached.

Educational analysis for authorized security professionals. Use only in controlled, authorized environments.

Sponsored
Share articleFacebookXLinkedIn

Continue exploring

Browse all articles →

0 comentários:

Post a Comment

Note: Only a member of this blog may post a comment.