Threat reportPhishingTL-2026-2537
N0va Phishkit Uses Device Code Phishing to Bypass MFA and Hijack SSO Sessions Across US and EU
The N0va phishing kit impersonates trusted platforms (Microsoft Teams, SharePoint, OneDrive, DocuSign, Google Drive, Dropbox, Zoom, Adobe Sign) to guide victims through the legitimate OAuth 2.0 Device Authorization Grant flow, capturing access/refresh tokens after MFA completes and exchanging them for Primary Refresh Token-based SSO access.
Key facts
| Identifiers | TL-2026-2537 |
|---|---|
| Also known as | Nova Phishkit |
| Severity | HIGH |
| Affected products | Microsoft Microsoft Entra ID / Microsoft 365 (OAuth 2.0 Device Authorization Grant / "device code" flow); Multiple (impersonated only, not themselves vulnerable) Microsoft Teams, SharePoint, OneDrive, DocuSign, Google Drive, Dropbox, Zoom, Adobe Sign |
| Tools | N0va |
| MITRE ATT&CK techniques | T1036.005 Match Legitimate Resource Name or Location, T1078.004 Cloud Accounts, T1090.002 External Proxy, T1098.005 Device Registration, T1213 Data from Information Repositories, T1528 Steal Application Access Token, T1550.001 Application Access Token, T1556 Modify Authentication Process, T1566.002 Spearphishing Link, T1583.006 Web Services, T1584.004 Server, T1657 Financial Theft |
| Motivation | FINANCIAL |
| Target sectors | government administration, technology, consulting, health |
| Target regions | North America, Europe |
| Detection rules | 9 (1 shown free) |
| Sources | 7 cited sources |
Detection coverage
As of 2026-09-16, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2537 across Splunk SPL, Microsoft KQL and Sigma, covering 14 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.
Free detection: N0va Device-Code Phishing Initiation — Proxy/SWG URL Signature & Brand-Lure Correlation
Detects the exploitation phase of the N0va phishkit: web/proxy traffic hitting the distinguishing backend URL signature (`/api/verification/init?session=*&flow=*prompt_profile=`) identified by ANY.RUN Threat Intelligence Lookup, or traffic to N0va's split hosting (Cloudflare Workers *.workers.dev, Linode Object Storage) combined with a brand-impersonation lure path (Teams, SharePoint, OneDrive, DocuSign, Google Drive, Dropbox, Zoom, Adobe Sign) — the moment a victim is walked toward the legitimate OAuth 2.0 Device Authorization Grant flow.
- Languages: Splunk SPL, Microsoft KQL, Sigma
- ATT&CK: T1566.002, T1036.005, T1584.004, T1583.006
- Data sources: proxy/SWG logs (Zscaler, Palo Alto, Squid, Blue Coat), web_gateway, azuread_signin_logs
- Severity: high
- Confidence: high
Splunk SPL
(index=proxy OR index=web) (sourcetype=zscaler OR sourcetype=pan:threat OR sourcetype=squid OR sourcetype=bluecoat)
| eval n0va_uri_match=if(match(url, "(?i)\/api\/verification\/init\?session=.*flow=.*prompt_profile="), 1, 0)
| eval n0va_infra_match=if(match(dest_host, "(?i)\.workers\.dev$") OR match(dest_host, "(?i)\.linodeobjects\.com$"), 1, 0)
| eval brand_lure_match=if(match(url, "(?i)(teams|sharepoint|onedrive|docusign|drive\.google|dropbox|zoom|adobesign)"), 1, 0)
| where n0va_uri_match=1 OR (n0va_infra_match=1 AND brand_lure_match=1)
| stats count earliest(_time) as first_seen latest(_time) as last_seen values(url) as urls values(dest_host) as hosts by src_ip, user
| eval first_seen=strftime(first_seen,"%Y-%m-%d %H:%M:%S"), last_seen=strftime(last_seen,"%Y-%m-%d %H:%M:%S")
| where count > 0Microsoft KQL
let deviceCodeSignins = SigninLogs
| where TimeGenerated > ago(24h)
| where AuthenticationProtocol == "deviceCode"
| project UserPrincipalName, SigninTime = TimeGenerated, IPAddress, AppDisplayName, Location;
let n0vaSignatureHits = CommonSecurityLog
| where TimeGenerated > ago(24h)
| where RequestURL has "/api/verification/init" and RequestURL has "prompt_profile="
| project SourceIP, HitTime = TimeGenerated, RequestURL;
deviceCodeSignins
| join kind=leftouter (n0vaSignatureHits) on $left.IPAddress == $right.SourceIP
| project SigninTime, UserPrincipalName, IPAddress, AppDisplayName, Location, HitTime, RequestURL
| where isnotempty(RequestURL) or isnotempty(UserPrincipalName)Sigma
title: N0va Phishkit Device-Code Verification API Signature
id: a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d
status: experimental
description: Detects the N0va phishing kit's distinguishing backend URL signature used during OAuth device-code phishing flows, as identified by ANY.RUN Threat Intelligence Lookup.
references:
- https://thehackernews.com/2026/09/n0va-phishkit-targets-us-and-eu.html
- https://x.com/anyrun_app/status/2095142285486821549
author: Threadlinqs Team
date: 2026/09/16
tags:
- attack.initial-access
- attack.t1566.002
- attack.defense-evasion
- attack.t1036.005
- attack.resource-development
- attack.t1584.004
- attack.t1583.006
logsource:
category: proxy
product: network
detection:
selection_uri:
cs-uri-query|contains: '/api/verification/init'
selection_params:
cs-uri-query|contains|all:
- 'session='
- 'flow='
- 'prompt_profile='
condition: selection_uri and selection_params
falsepositives:
- Unknown; this exact URL parameter pattern has only been observed on N0va infrastructure per ANY.RUN reporting
level: highKnown false positives:
- Legitimate internal tooling or security research that happens to reuse the same URL query-parameter names (session/flow/prompt_profile) on unrelated infrastructure — extremely unlikely but not provably impossible
Tuning notes: Maintain an allowlist of known-legitimate *.workers.dev / *.linodeobjects.com tenants your org actually uses (e.g. internal serverless apps) to suppress the infra-only leg of the match; the URL-signature leg alone is high-fidelity and should not be suppressed.
- N0va Device-Code Grant Sign-In Detection — Entra ID SigninLogs · unlock with TL Creds
- Entra ID Device Registration Post Device-Code Grant — SSO Persistence · unlock with TL Creds
- Sentinel: Token-Reuse-Driven SharePoint/OneDrive/Mailbox Access Following N0va Infra Contact · unlock with TL Creds
- N0va Backend Verification-API URL Signature (Proxy/Web) · unlock with TL Creds
- Device-Code Sign-In Chained to Device Registration (Sigma Correlation) · unlock with TL Creds
- N0va Relay Infrastructure Contact — Cloudflare Workers / Linode Object Storage · unlock with TL Creds
- Device Registration Immediately Following Device-Code Sign-In (PRT / SSO Persistence Abuse) · unlock with TL Creds
- Stolen OAuth Token Reuse for SSO / Data Access via N0va Relay Infrastructure · unlock with TL Creds
Indicators of compromise
As of 2026-09-16, Threadlinqs Intelligence tracks 14 indicators of compromise for TL-2026-2537.
Sample indicators (1 of 14)
- URL:
hxxps://microsoft[.]com/devicelogin
Defanged, TLP:CLEAR. Full set of 14 indicators: unlock with TL Creds.
Technical analysis
It has hit government, technology, consulting, and healthcare organizations across North America and Europe, with confirmed payment fraud, invoice manipulation, and data exposure.
N0va is a phishing kit disclosed by ANY.RUN researchers in September 2026 that abuses Microsoft's OAuth 2.0 Device Authorization Grant ("device code") flow to obtain valid access and refresh tokens without ever presenting a credential-harvesting form or a look-alike login domain. Victims are lured with pages impersonating eight widely trusted business platforms -- Microsoft Teams, SharePoint, OneDrive, DocuSign, Google Drive, Dropbox, Zoom, and Adobe Sign -- and are instructed to browse to the real Microsoft device-login portal and enter an attacker-generated user code. Because the victim authenticates on Microsoft's own infrastructure and completes their normal MFA challenge, the attacker's device-code session is authorized as if it were the victim's own device, yielding usable OAuth tokens even though MFA was satisfied.
The documented attack chain runs: trusted-brand lure -> device-code phishing -> legitimate authentication -> access/refresh token capture -> token exchange and device registration -> SSO access to corporate resources. Once tokens are captured, N0va operators exchange them and abuse device-registration mechanisms to obtain Primary Refresh Token (PRT)-based single sign-on access, allowing follow-on activity to blend in with normal authenticated account use rather than triggering malware or anomalous-login detections.
N0va's infrastructure is deliberately split across compromised legitimate websites, Cloudflare Workers subdomains (*.workers.dev), and Linode Object Storage, which spreads indicators across domain reputation, hosting, and authentication telemetry so that no single detection layer sees the full chain. ANY.RUN's Threat Intelligence Lookup identified a distinguishing backend URL signature, `/api/verification/init?session=*&flow=*prompt_profile=`, that has been used to pivot onto additional related N0va infrastructure. ANY.RUN's interactive sandbox reproduced complete N0va attack chains in roughly 24 seconds, underscoring how quickly the multi-stage flow completes once a victim engages.
Observed targeting spans government, technology, consulting, and healthcare organizations in North America and the European Union. Confirmed operational impact includes payment fraud and invoice manipulation, and exposure of customer records, employee data, and intellectual property following SSO-scoped access to email, file-sharing, and collaboration platforms. No CVE or CVSS score applies -- this is an abuse of a legitimate, by-design OAuth authentication flow rather than a software vulnerability -- and no source attributes N0va to a specific named threat actor or nation-state.
N0va is the latest entrant in a device-code-phishing tradecraft wave that Microsoft first formalized publicly in February 2025 with its disclosure of Storm-2372 (active since roughly August 2024, targeting government, NGO, IT, defense, telecom, health, and energy organizations across Europe, North America, Africa, and the Middle East, assessed with moderate confidence to align with Russian interests). The Cloud Security Alliance documented the technique's continued proliferation at scale in a March 2026 research note covering more than 340 affected Microsoft 365 organizations, and open reporting references a separate April 2026 Microsoft-documented campaign using identical device-code-authentication weaknesses, as well as a distinct "BigBear 2.0" session-hijacking campaign against Microsoft 365 tenants. N0va's disclosure indicates device-code phishing has matured from a single-actor technique into repeatable, actor-agnostic phishkit tradecraft.
MITRE ATT&CK techniques by tactic:
Defense Evasion
T1036.005 Match Legitimate Resource Name or Location
Persistence
T1078.004 Cloud Accounts; T1098.005 Device Registration
Command and Control
Collection
T1213 Data from Information Repositories
Credential Access
T1528 Steal Application Access Token; T1556 Modify Authentication Process
lateral-movement
T1550.001 Application Access Token
Initial Access
Resource Development
T1583.006 Web Services; T1584.004 Server
Impact
Remediation
Immediate actions
- Restrict or disable the OAuth 2.0 Device Authorization Grant (device code flow) tenant-wide via Microsoft Entra Conditional Access unless a specific line-of-business need requires it
- Revoke refresh and access tokens and force re-authentication for any account that completed a device-code sign-in from an unfamiliar application, client ID, geography, or ASN
- Hunt Entra ID / Microsoft 365 sign-in logs for the device-code grant type combined with unfamiliar client IDs or immediately-following device-registration events
- Block or alert on the identified N0va URL signature `/api/verification/init?session=*&flow=*prompt_profile=` and on referrers from Cloudflare Workers (*.workers.dev) and Linode Object Storage at the web proxy / SWG
Workarounds
- Where the device code flow cannot be disabled outright, restrict it to specific named applications and trusted network locations using Conditional Access named locations and application-based Conditional Access
Longer-term hardening
- Move high-value users to phishing-resistant MFA (FIDO2/WebAuthn, certificate-based authentication) that cannot be satisfied by simply approving a displayed code
- Require compliant/managed devices via Conditional Access for any flow that still permits device-code or legacy authentication
- Add SSL/TLS decryption and identity-aware web proxy inspection so device-code phishing pages and follow-on API calls are visible to network security tooling
- Correlate identity-provider sign-in and token-issuance telemetry with email, endpoint, and network detections in SIEM/SOAR rather than relying on any single signal
Affected and fixed versions
- Microsoft — Microsoft Entra ID / Microsoft 365 (OAuth 2.0 Device Authorization Grant / "device code" flow)
Vulnerable versions: Tenants with the device code authentication flow enabled by default and no restricting Conditional Access policy
Fixed in: Tenants with Conditional Access policies that block or restrict the device-code flow to trusted apps/locations - Multiple (impersonated only, not themselves vulnerable) — Microsoft Teams, SharePoint, OneDrive, DocuSign, Google Drive, Dropbox, Zoom, Adobe Sign
Timeline
- Microsoft Threat Intelligence later reports that Storm-2372 began conducting device-code phishing campaigns around this time, targeting government, NGO, IT, defense, telecom, health, and energy organizations across Europe, North America, Africa, and the Middle East -- establishing the device-code-authentication-abuse tradecraft that N0va would later reuse. (Related precedent activity, not N0va itself.)
- Microsoft Security Blog publishes "Storm-2372 conducts device code phishing campaign," publicly formalizing detection and defense guidance for the OAuth device-code-phishing technique that N0va later leverages. (Background reference, not N0va-specific.)
- The Cloud Security Alliance publishes a research note documenting device-code phishing affecting more than 340 Microsoft 365 organizations, showing the technique had proliferated well beyond a single actor ahead of N0va's disclosure. (Background reference, not N0va-specific.)
- Open reporting references a separate Microsoft-documented campaign employing identical device-code-authentication weaknesses, indicating continued active abuse of the technique by multiple operators. (Related campaign, not N0va itself.)
- ANY.RUN researchers disclose the N0va phishkit, identifying its brand-impersonation lures, its device-code-phishing attack chain, its split infrastructure across compromised sites, Cloudflare Workers, and Linode Object Storage, and a distinguishing backend URL signature used to pivot onto related infrastructure via Threat Intelligence Lookup.
- eSecurityPlanet, Cybersecurity News, and First Hackers News publish follow-on analyses of N0va, adding detection guidance covering interactive sandboxing, cross-signal correlation across domain/IP/hosting/authentication activity, and SIEM/SOAR/EDR integration.
- The Hacker News publishes coverage of N0va based on ANY.RUN's findings, detailing targeting of government, technology, consulting, and healthcare organizations in North America and the EU and confirming operational impact including payment fraud, invoice manipulation, and exposure of customer records, employee data, and intellectual property.
Sources cited
- ANY.RUN: We uncovered #N0va, a new phishkit targeting organizations in North America and Europe x.com · Research · 2026-09-16
- Storm-2372 conducts device code phishing campaign Microsoft · Vendor report · 2025-02-13
- OAuth Device Code Phishing Hits 340+ Microsoft 365 Organizations labs.cloudsecurityalliance.org · Research · 2026-03-25
- N0va Phishkit Targets US and EU The Hacker News · News · 2026-09-16
- N0va Phishkit Targets North America and Europe Through Microsoft Logins esecurityplanet.com · News · 2026-09-16
- New N0va Phishkit Targets North America and EU: A Growing Identity Risk for SOCs cybersecuritynews.com · News · 2026-09-16
- New N0va Phishkit Campaign Puts Enterprise Identities At Risk firsthackersnews.com · News · 2026-09-16
Previous and next in phishing
Drafted by the Threadlinqs research pipeline from 7 sources; published ; last updated —.