TeamFiltration Returns: UNK_CondorFiltration Credential-Spraying Campaign Targets Dormant M365 Service Accounts in Latin America

TeamFiltration Returns (TL-2026-2616), also tracked as Spraying in the Andes, is a high-severity cloud security threat, first published 2026-09-22. It is attributed to UNK_CondorFiltration with low confidence, affects Microsoft Microsoft 365 / Entra ID, maps to 13 MITRE ATT&CK techniques (T1036, T1078.004, T1087.004), and is covered by 9 detection rules and 7 indicators of compromise.

Key facts for TL-2026-2616

Threat ID
TL-2026-2616
Also known as
Spraying in the Andes
Severity
HIGH
Status
ACTIVE
Category
CLOUD
First published
2026-09-22
Last reviewed
2026-09-22
Attribution
UNK_CondorFiltration
Attribution confidence
LOW
Motivation
UNKNOWN
Target sectors
retail, finance
Target regions
Latin America, chile
Detection rules
9
Indicators of compromise
7

Malware and tooling in TeamFiltration Returns

Malware and tooling: FireProx, TeamFiltration

Proofpoint identified a renewed campaign, tracked as UNK_CondorFiltration, using the open-source TeamFiltration offensive framework to credential-spray 28 Microsoft 365 tenants across Latin America, concentrated on Chilean retail and financial-sector organizations. Over 5,714 accounts were targeted across three waves between July 21 and August 16, 2026, resulting in 7 confirmed compromises of dormant, unmanaged service accounts lacking MFA, followed by post-compromise reconnaissance and attempted exfiltration of email, Teams chats, and OneDrive/SharePoint data.

How TeamFiltration Returns works

Between July 21 and August 16, 2026, Proofpoint tracked a three-wave password-spraying campaign against 28 Microsoft 365 tenants concentrated in Chile, attributing the activity to a cluster it designates UNK_CondorFiltration. The campaign generated 32,825 authentication events against 5,714 unique accounts from 1,487 unique source IPs, all resolving to Amazon Web Services (AWS) EC2 ranges (3.101.0.0/16, 18.144.76.0/24, 13.52.201.0/24). Proofpoint assesses the actor rotated across AWS regions via the FireProx IP-rotation technique to evade IP-based blocking during both the enumeration and spray phases.

The activity is consistent with TeamFiltration, a cross-platform (.NET) offensive framework for enumerating, spraying, exfiltrating from, and backdooring Microsoft 365 / Entra ID (formerly Azure AD) accounts. TeamFiltration was built internally at TrustedSec by Joakim Kandefelt (@Flangvik) starting in January 2021 and publicly released at DEF CON 30. Its enumeration module validates account existence via the Microsoft Teams API at roughly 300 accounts/second without triggering lockouts; its spray module tests passwords against the resulting validated-account list; its exfiltration module pulls Graph API user/group data, the last ~2,000 Outlook emails via the Outlook REST API, and Teams/OneDrive/SharePoint files and chats; and an interactive backdoor module can grant persistent OneDrive access. All observed requests carried a single hardcoded, outdated Microsoft Teams Electron client User-Agent string (Teams/1.3.00.30866, Chrome/80.0.3987.165, Electron/8.5.1) — the campaign's most reliable detection fingerprint.

Campaign activity ran in three distinct waves: July 21-24 (roughly 100-120 accounts/day against Chilean banks), July 26-28 (peaking at ~1,520 accounts on July 27 against a financial institution), and August 13-16 (peaking at ~1,560 accounts on August 15 against a major, unnamed Chilean retailer that alone accounted for 53.2% of all targeted accounts and 78.3% of all authentication events in the campaign). Every one of the 7 confirmed compromises traced to the same root cause: a default, unrotated password on a dormant or unmanaged service/functional account with no MFA enforcement and no prior sign-in history. Proofpoint observed the actor pivot from AWS-sourced initial access to the Azure Portal (triggering MFA enrollment prompts) within approximately 90 seconds, switch to a German VPN egress node (149.88.104.19, hosted via cdn77.com) for follow-on activity, conduct SharePoint Online reconnaissance, request Microsoft Graph API tokens, and attempt (unsuccessfully) to probe corporate VPN portals — those attempts were blocked by MFA and Conditional Access policies that were not applied to the compromised service accounts themselves. Applications accessed included Microsoft Teams (client ID 1fec8e78-bce4-4aaf-ab1b-5451cc387264), Office/Outlook (d3590ed6-52b3-4102-aeff-aad2292ab01c), OneDrive (ab9b8c07-8f02-4f72-87fa-80105867a763), Azure Portal (c44b4083-3bb0-49c1-b47d-974e53cbdf3c), and SharePoint (00000003-0000-0ff1-ce00-000000000000).

TeamFiltration abuse is not new: Proofpoint separately tracked a distinct, related activity set, UNK_SneakyStrike, that used the same tool to target over 80,000 Entra ID accounts across roughly 100 cloud tenants beginning in December 2024 (peaking January 8, 2025 at 16,500 accounts in a single day), sourced primarily from US/Ireland/UK AWS regions, exploiting Microsoft OAuth 'family refresh token' abuse for broader account access and exhibiting the same distinctive outdated Teams user-agent. UNK_CondorFiltration's AWS-EC2/FireProx tradecraft and identical tool fingerprint are consistent with that lineage, though Proofpoint tracks it as a separate cluster given the differing infrastructure, targeting geography (Latin America vs. global), and campaign timing. The recurring structural gap across both clusters is the same: forgotten, unmanaged, non-interactive service accounts are provisioned without MFA and go unmonitored, making them the most exploitable identity in an otherwise hardened M365 tenant.

MITRE ATT&CK techniques used in TL-2026-2616

Stealth

T1036 Masquerading

Initial Access

T1078.004 Cloud Accounts

Persistence

T1078.004 Cloud Accounts

Discovery

T1087.004 Cloud Account; T1526 Cloud Service Discovery

Command and Control

T1090.002 External Proxy

Credential Access

T1110.003 Password Spraying; T1528 Steal Application Access Token

Collection

T1114.002 Remote Email Collection; T1119 Automated Collection; T1213 Data from Information Repositories

Exfiltration

T1567 Exfiltration Over Web Service

Resource Development

T1583.006 Web Services

Reconnaissance

T1589 Gather Victim Identity Information

Affected products and versions in TeamFiltration Returns

  • Microsoft — Microsoft 365 / Entra ID
    Vulnerable versions: Tenants with dormant or unmanaged service/functional accounts lacking MFA and Conditional Access enforcement
    Fixed in: Tenants enforcing MFA and Conditional Access uniformly across all accounts, including service accounts

Remediation for TeamFiltration Returns

Immediate actions

  • Enforce MFA and Conditional Access on all Microsoft 365 / Entra ID accounts, including dormant and unmanaged service/functional accounts
  • Audit all service accounts tenant-wide for default or unrotated passwords and rotate/disable any with no recent legitimate sign-in history
  • Alert on and block authentication attempts to service accounts originating from AWS EC2 IP ranges (3.101.0.0/16, 18.144.76.0/24, 13.52.201.0/24) or other unexpected cloud-hosting ASNs

Workarounds

  • Hunt Entra ID / M365 unified audit sign-in logs for the hardcoded TeamFiltration Teams client User-Agent (Teams/1.3.00.30866 Chrome/80.0.3987.165 Electron/8.5.1)
  • Flag sign-ins to Teams, Office, OneDrive, Azure Portal, or SharePoint (application IDs 1fec8e78-bce4-4aaf-ab1b-5451cc387264, d3590ed6-52b3-4102-aeff-aad2292ab01c, ab9b8c07-8f02-4f72-87fa-80105867a763, c44b4083-3bb0-49c1-b47d-974e53cbdf3c, 00000003-0000-0ff1-ce00-000000000000) from devices/user-agents inconsistent with the signing-in account's normal client
  • Correlate concentrated, wave-shaped bursts of failed authentication from sequential/rotating source IPs against many distinct accounts as a password-spray indicator

Longer-term hardening

  • Establish and monitor sign-in baselines per service account (expected source ASN/geo, time window, application) and alert on deviation
  • Inventory and lifecycle-manage all non-interactive/service accounts across every M365 tenant, including deprovisioning genuinely unused accounts
  • Require Conditional Access policies (compliant device, MFA, trusted location) apply uniformly to service accounts, not just interactive user accounts

Timeline of TeamFiltration Returns

  • Wave 1 begins: UNK_CondorFiltration credential-sprays Chilean banks at roughly 100-120 accounts/day from AWS EC2-sourced infrastructure using TeamFiltration-consistent tooling.
  • Wave 1 of the campaign concludes.
  • Wave 2 begins, shifting focus to a Chilean financial institution.
  • Wave 2 peaks at approximately 1,520 accounts targeted in a single day.
  • Wave 2 of the campaign concludes.
  • Wave 3 begins, targeting a major, unnamed Chilean retailer that ultimately accounts for 53.2% of all targeted accounts and 78.3% of all authentication events in the campaign.
  • Wave 3 peaks at approximately 1,560 accounts targeted in a single day.
  • Campaign activity window closes. Final totals: 5,714 unique accounts targeted across 28 M365 tenants, 32,825 authentication events from 1,487 unique AWS EC2 source IPs, and 7 confirmed compromises of dormant/unmanaged service accounts. Post-compromise activity on the confirmed accounts included an AWS-to-Azure-Portal pivot within roughly 90 seconds of initial access, a switch to a German VPN egress node (149.88.104.19 via cdn77.com), SharePoint Online reconnaissance, Microsoft Graph API token requests, and blocked attempts to probe corporate VPN portals (stopped by MFA/Conditional Access).
  • Proofpoint Threat Insight publishes 'Spraying in the Andes: TeamFiltration Returns to Exploit Forgotten Service Accounts,' publicly disclosing the UNK_CondorFiltration campaign and attributing it to the TeamFiltration framework.

Sources cited for TeamFiltration Returns

More in cloud

Detection coverage for TL-2026-2616

As of 2026-09-22, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2616 across Splunk SPL, Microsoft KQL and Sigma, covering 7 indicator(s) of compromise. The whole corpus is readable without an account; a free account unlocks full detection query text in Splunk SPL, Microsoft KQL and Sigma; paid tiers add raw indicator values, correlation and the MCP server. Threadlinqs MCP server · View plans.

Threadlinqs Intelligence — Real-Time Threat Detection Platform

[ 0 threats ] [ 0 det ] [ CRIT: 0 ] [ HIGH: 0 ]
// threat_feed
$ sort --newest
Showing all threats

Latest Threats