
Summary
This rule detects a successful Microsoft Entra ID sign-in where the browser family and operating system family combination has not previously appeared for the user within the history window, and where the authentication did not complete a phishing-resistant method and did not originate from an Entra-joined or registered device. It targets adversary‑in‑the‑middle proxies or automated sign-ins that spoof a client fingerprint to suppress the passkey prompt and force a phishable fallback. The detection is implemented as a new_terms rule that flags a novel combination of user_agent.name, azure.signinlogs.properties.device_detail.operating_system, and user.name within the configured window (now-9m historically) and corroborates that the session used a non‑phishing‑resistant method, with no trusted device in device_detail.trust_type. The rule maps to MITRE ATT&CK technique T1556 (Modify Authentication Process) with subtechnique T1556.006 (MFA) and is categorized under Credential Access. It relies on the Entra ID Sign-In Logs data stream (Azure Event Hub integration) and notes that impossible pairings (e.g., Safari on Windows) may indicate a spoofed fingerprint used to bypass stronger authentication.false_positives discuss legitimate but rare client changes, privacy/proxy interference, and the need to baseline and whitelist sanctioned clients. The rule includes actionable triage steps (analyze new combination against history, verify source IP/ASN/geo, inspect authentication_details for downgraded methods, and check conditional_access_status) and remediation guidance (revoke active sessions, refresh tokens, remove attacker-registered devices, reset credentials, and enforce phishing-resistant CA to prevent MFA downgrades). References to MFA-downgrade and phishing-resistance are provided, along with a risk score (47) and a medium severity. The New Terms mechanism expands detection on fields: user.name, user_agent.name, and azure.signinlogs.properties.device_detail.operating_system within a history window (now-10d) to identify truly novel client configurations, with correlation to subsequent token reuse as an indicator of compromised sessions.
Categories
- Cloud
- Identity Management
- Web
Data Sources
- Cloud Service
- Web Credential
- Logon Session
- Application Log
ATT&CK Techniques
- T1556
- T1556.006
Created: 2026-10-21