As reported by BleepingComputer, Huntress researchers have published detailed analysis of three modern phishing kits — EvilTokens, Knight Office, and Kali365 — purpose-built to steal Microsoft 365 session tokens. The findings underscore a shift that security teams have been tracking for years but many organizations still haven't operationalized: the attacker's target has moved from credentials to sessions, and traditional MFA is no longer the stopping point it was once assumed to be.
The Core Problem: Sessions > Passwords
The Huntress analysis highlights three tradecraft techniques — device code phishing, Adversary-in-the-Middle (AiTM) relay, and direct token theft — all converging on the same objective: obtaining a valid, authenticated session that has already passed MFA. This is significant because most organizations still treat successful MFA completion as a trust signal. It isn't. A completed MFA challenge tells you the user authenticated at some point; it tells you nothing about whether that session is now controlled by the legitimate user or an attacker holding a stolen token.
The commodification angle is equally important. These kits ship with admin dashboards, real-time victim tracking, and support channels — the operational maturity of a SaaS product. This lowers the barrier to entry so that even unsophisticated actors can execute what was once advanced tradecraft. The threat surface isn't just well-funded groups; it's anyone with a kit and a lure.
Why This Matters Now
Microsoft 365 is the identity backbone for the majority of enterprises. A stolen session token grants access to email, files, Teams, and — critically — the ability to establish persistence through OAuth app registrations, mail forwarding rules, and additional account creation. Initial access via M365 session theft is now a common precursor to Business Email Compromise (BEC), data exfiltration, and lateral movement into on-premises or hybrid environments.
What Defenders Should Focus On
The defensive response cannot be "make phishing harder" — it must be "make stolen sessions worthless." That requires a fundamentally different control posture:
Detection Gaps Most Teams Still Have
Most SIEM detections are still built around authentication events — failed logins, impossible travel, MFA fatigue. But session token replay often produces no authentication event at all, because the attacker is using a valid token. Teams need detections focused on post-authentication behavior: unusual mail forwarding rules, bulk downloads from OneDrive/SharePoint, new OAuth consent, and API access patterns that deviate from the user's baseline.
Microsoft's Unified Audit Log (UAL) remains underutilized. The MailItemsAccessed and MailboxLogin events in particular can reveal session-based access that bypasses traditional sign-in logs.
Shield53 Recommendations
- Enable Entra ID Continuous Access Evaluation (CAE) — this is the single most impactful control against token replay. CAE enforces policy re-evaluation during active sessions, not just at sign-in.
- Implement session lifetime policies with shorter token validity windows and stricter refresh token conditions for high-risk users (administrators, finance, executives).
- Deploy or tune sign-in risk policies in Identity Protection with automated remediation (require re-authentication, block, or MFA) for medium-and-above risk levels.
- Audit OAuth app registrations weekly — query Graph API for newly registered apps and consent grants; alert on any app with
Mail.ReadorMail.ReadWritescope. - Train help desk and SOC on session-based indicators of compromise — new inbox rules, unexpected forwarding, simultaneous sessions from disparate geographies.
- Consider FIDO2 / phishing-resistant MFA — while not a silver bullet against all token theft, passkeys eliminate the AiTM relay path by binding authentication to the origin domain.
The defensive mindset must shift from "did the user authenticate correctly?" to "is this session behaving like the legitimate user?" Token theft means the first question is no longer sufficient.
The Huntress report is a useful reminder that the adversary's tooling has outpaced many organizations' detection capabilities. The kits are commodified; the defenses must be too — standardized, automated, and built around the assumption that initial access will succeed and the real fight begins after.