JPMorgan Chase has confirmed a customer data breach traced back to a compromise at a third-party vendor. An extortion actor operating under the handle N0n has claimed responsibility not only for the JPMorgan intrusion but also for attacks against Visa and Mastercard, and is reportedly advertising stolen customer data on its dark web leak site. This follows the now-standard double-extortion playbook: infiltrate a softer vendor in the supply chain, exfiltrate data, encrypt where possible, and pressure the downstream brand-name victim with public leaks.
For defenders, the details that matter are not the headlines — they are the mechanics. The initial access vector was not JPMorgan's own perimeter. It was a trusted third party with legitimate connectivity, credentials, or data access into the financial giant's environment. That is the defining feature of modern supply-chain and third-party compromises, and it is exactly the blind spot most detection programs still have. If your SOC only monitors your own assets and treats vendor access as inherently trusted, you are carrying the same risk that just burned one of the most heavily defended institutions on the planet.
This post breaks down the incident from a defensive perspective: what the attack pattern looks like, how to hunt for the same TTPs in your own environment, and what to remediate now — whether you are a financial institution, a vendor to one, or any organization with third-party access into your network.
Technical Analysis
What We Know
- Threat actor: N0n, an extortion-focused group described as running encryption-based cyber incidents and operating a dark web leak site.
- Primary confirmed victim: JPMorgan Chase — customer data exposed via a third-party compromise.
- Additional claimed victims: Visa and Mastercard (claims, as of reporting, not fully independently confirmed).
- Exfiltration/extortion: Customer data reportedly leaked on the actor's dark web site, consistent with data-theft-extortion operations where leak publication is the leverage.
- Initial vector: A third-party vendor compromise — not a direct breach of JPMorgan's core infrastructure.
No CVE has been publicly associated with this incident, and no specific vendor product has been named as the exploited component. Do not chase a vulnerability identifier here — the defensive lesson is architectural, not patch-based.
The Attack Chain (Defender's View)
Based on the reported structure of this incident and the established playbook of extortion actors abusing third-party access, the kill chain typically looks like this:
- Vendor compromise (T1199 – Trusted Relationship): The actor gains access to a vendor's environment — commonly via stolen credentials, an exposed remote access service, or phishing against vendor staff. The vendor holds legitimate credentials, API keys, VPN access, or data-sharing pipelines into the target.
- Abuse of trusted access (T1078 – Valid Accounts): Rather than exploiting the target directly, the actor rides the vendor's sanctioned access path. This traffic looks legitimate to perimeter controls because, technically, it is — until behavior deviates.
- Discovery and collection (T1083, T1560): The actor enumerates accessible shares, databases, and storage (often the exact repositories the vendor integration was built to touch), then stages data — frequently into compressed/encrypted archives in temp or staging directories.
- Exfiltration (T1048 / T1567): Bulk outbound transfer to actor-controlled infrastructure, commonly over HTTPS to cloud storage or dedicated exfil servers, timed to blend with business traffic.
- Extortion (T1657 – Financial Theft / TA0040 Impact): Leak site publication and pressure. Encryption may or may not occur — in pure extortion plays, data theft alone is the payload.
Exploitation Status
This is a confirmed, active incident with confirmed data exposure at JPMorgan. The actor's claims against Visa and Mastercard remain under investigation, but the operational pattern — hitting interconnected financial-sector entities through shared third parties — means every organization in these ecosystems should treat this as an active threat to their own vendor relationships, not a spectator event.
The critical takeaway: your detection surface must extend to third-party identities and integration paths. If a vendor account authenticates at 3:12 AM from an unfamiliar ASN and starts touching repositories it has never accessed before, that has to be a high-fidelity alert, not a log line.
Detection & Response
The detections below target the observable behaviors common to third-party-abuse extortion intrusions: anomalous vendor account activity, bulk data staging, and large-scale exfiltration. Tune thresholds to your environment before production deployment.
Sigma Rules
---
title: Anomalous Authentication from Third-Party or Vendor Account Outside Baseline
id: 8c2f4a61-3b7d-4e59-9a1c-2d6e8f0b4a73
status: experimental
description: Detects logons from accounts designated as third-party/vendor/service accounts originating from unusual source IPs, at unusual hours, or after long dormancy — a hallmark of trusted-relationship abuse as seen in the JPMorgan third-party breach.
references:
- https://attack.mitre.org/techniques/T1199/
- https://attack.mitre.org/techniques/T1078/
- https://www.infosecurity-magazine.com/news/jpmorgan-data-breach-third-party/
author: Security Arsenal
date: 2026/04/06
tags:
- attack.initial_access
- attack.t1199
- attack.persistence
- attack.t1078
logsource:
category: authentication
product: windows
detection:
selection_vendor_account:
TargetUserName|contains:
- 'svc-vendor'
- 'vendor-'
- 'partner-'
- 'thirdparty'
- 'ext-'
selection_unusual_source:
IpAddress|contains:
- '10.'
- '192.168.'
- '172.16.'
filter_known_vendor_nets:
IpAddress|startswith:
- '10.50.' # Replace with documented vendor VPN ranges
condition: selection_vendor_account and selection_unusual_source and not filter_known_vendor_nets
falsepositives:
- Vendor staff working from new locations — maintain an allowlist of documented vendor egress ranges and review quarterly
- Newly onboarded vendors not yet added to the allowlist
level: high
---
title: Bulk Data Staging with Archive Creation in Temp or Staging Directories
id: 4d9e1b72-8a3c-4f67-b2d5-7e1a9c0f5b28
status: experimental
description: Detects creation of compressed archives by non-administrative tooling in temp, staging, or user profile directories — consistent with pre-exfiltration data staging observed in extortion intrusions.
references:
- https://attack.mitre.org/techniques/T1560/001/
- https://www.infosecurity-magazine.com/news/jpmorgan-data-breach-third-party/
author: Security Arsenal
date: 2026/04/06
tags:
- attack.collection
- attack.t1560.001
logsource:
category: process_creation
product: windows
detection:
selection_tool:
Image|endswith:
- '\7z.exe'
- '\7za.exe'
- '\rar.exe'
- '\winrar.exe'
- '\tar.exe'
- '\makecab.exe'
selection_args:
CommandLine|contains:
- ' a '
- ' -r'
- ' -p'
- 'u -r'
selection_path:
CommandLine|contains:
- '\Temp\'
- '\AppData\'
- '\ProgramData\'
- 'C:\Users\Public\'
- '\staging\'
filter_admin:
User|contains: 'svc-backup'
condition: selection_tool and selection_args and selection_path and not filter_admin
falsepositives:
- Software packaging and deployment activity — scope to servers hosting vendor-accessible data
- IT support tools creating archives for log collection
level: medium
---
title: High-Volume Outbound Data Transfer from Server Hosting Third-Party Integrations
id: 1f7c5d93-6b4a-4e28-c3f9-8d2b6a1e4c07
status: experimental
description: Detects abnormally large outbound network transfers from servers that host vendor integrations, file shares, or data exchange services — a key exfiltration indicator in third-party-breach scenarios.
references:
- https://attack.mitre.org/techniques/T1048/
- https://attack.mitre.org/techniques/T1567/
author: Security Arsenal
date: 2026/04/06
tags:
- attack.exfiltration
- attack.t1048
- attack.t1567
logsource:
category: network_connection
product: windows
detection:
selection:
Initiated: 'true'
DestinationPort:
- 443
- 22
- 21
- 989
- 990
filter_internal:
DestinationIp|startswith:
- '10.'
- '192.168.'
- '172.16.'
- '172.17.'
- '172.18.'
- '172.19.'
- '172.2'
- '172.30.'
- '172.31.'
- '127.'
filter_known_backup:
DestinationIp|contains:
- '198.51.100.10' # Replace with sanctioned backup/DR endpoints
condition: selection and not filter_internal and not filter_known_backup
falsepositives:
- Sanctioned cloud backup and replication jobs — baseline volume per host and alert on deviation rather than raw events
- Software update distribution
level: medium
KQL — Microsoft Sentinel / Defender
This hunt correlates three behaviors across a 7-day window: vendor account authentication anomalies, archive-creation process execution, and large outbound transfers. Run it against environments with third-party integrations — especially financial services, payment processing, and any B2B data exchange.
let lookback = 7d;
let vendorAccounts = dynamic(["svc-vendor", "vendor-", "partner-", "thirdparty", "ext-"]);
// Stage 1: Vendor-account logons from sources not seen in the prior 30 days
let baseline = SecurityEvent
| where TimeGenerated between (ago(30d) .. ago(lookback))
| where EventID == 4624
| where Account has_any (vendorAccounts)
| summarize by Account, IpAddress;
let anomalousLogons = SecurityEvent
| where TimeGenerated >= ago(lookback)
| where EventID == 4624 and LogonType in (3, 10)
| where Account has_any (vendorAccounts)
| summarize arg_max(TimeGenerated, *) by Account, IpAddress
| where IpAddress !in (baseline | project IpAddress)
| project LogonTime=TimeGenerated, Account, IpAddress, Computer;
// Stage 2: Archive-creation tooling executed on vendor-accessible hosts
let staging = DeviceProcessEvents
| where TimeGenerated >= ago(lookback)
| where FileName in~ ("7z.exe", "7za.exe", "rar.exe", "winrar.exe", "tar.exe", "makecab.exe")
| where ProcessCommandLine has_any ("\\Temp\\", "\\AppData\\", "\\ProgramData\\", "Public")
| project StagingTime=TimeGenerated, DeviceName, AccountName, FileName, ProcessCommandLine;
// Stage 3: Large outbound transfers from the same hosts
let exfil = DeviceNetworkEvents
| where TimeGenerated >= ago(lookback)
| where RemotePort in (443, 22, 21, 990)
| where RemoteIPType == "Public"
| summarize TotalConnections=count(), DistinctDestinations=dcount(RemoteIP), Destinations=make_set(RemoteIP) by DeviceName, InitiatingProcessAccountName, bin(TimeGenerated, 1h)
| where DistinctDestinations > 20 or TotalConnections > 500;
// Correlate all three stages
anomalousLogons
| join kind=inner (staging) on $left.Computer == $right.DeviceName
| join kind=inner (exfil) on $left.Computer == $right.DeviceName
| project LogonTime, Account, IpAddress, Computer, StagingTime, FileName, ProcessCommandLine, TotalConnections, DistinctDestinations, Destinations
| sort by LogonTime asc
Velociraptor VQL
Deploy this artifact across servers that host vendor integrations, shared file repositories, or payment-data processing to identify staging and exfil indicators in a single sweep.
-- Hunt for extortion-style staging and exfil artifacts on vendor-integrated hosts
-- Part 1: Recently created large archives in staging locations
LET archives = SELECT FullPath, Size, Mtime, Ctime
FROM glob(globs=[
'C:/Windows/Temp/**/*.zip',
'C:/Windows/Temp/**/*.7z',
'C:/Windows/Temp/**/*.rar',
'C:/Users/Public/**/*.zip',
'C:/Users/Public/**/*.7z',
'C:/ProgramData/**/*.7z',
'C:/ProgramData/**/*.rar'
])
WHERE Size > 10000000
AND Mtime > timestamp(epoch=now() - 604800)
-- Part 2: Live processes with archive/exfil tooling in command line
LET tooling = SELECT Pid, Name, CommandLine, Exe, Username, CreateTime
FROM pslist()
WHERE CommandLine =~ '(7z|7za|rar|winrar|tar|makecab|rclone|megacmd|filezilla|winscp)'
-- Part 3: Outbound connections to public IPs from data-serving processes
LET connections = SELECT Pid, Name, RemoteAddress, RemotePort, Status
FROM netstat()
WHERE RemotePort in (443, 22, 21, 990)
AND RemoteAddress !~ '^(10\.|192\.168\.|172\.(1[6-9]|2[0-9]|3[01])\.|127\.)'
AND Status =~ 'ESTABLISHED'
SELECT 'archive' AS IndicatorType, FullPath AS Detail, format(format='%d MB', args=Size/1048576) AS Size, Mtime AS Timestamp, '' AS Process, '' AS Remote
FROM archives
UNION ALL
SELECT 'staging_process', CommandLine, '', CreateTime, Name, '' FROM tooling
UNION ALL
SELECT 'outbound_connection', RemoteAddress, '', NULL, Name, format(format='%v:%v', args=[RemoteAddress, RemotePort]) FROM connections
Remediation / Hardening Script
Use this PowerShell script to audit third-party and service accounts for the exact risk conditions this breach exploited: stale vendor accounts, unrestricted logon scope, and missing interactive-logon restrictions.
# Third-Party Account Audit & Hardening — run on a DC or admin workstation with RSAT
# Requires: ActiveDirectory module. Review output before disabling anything.
$report = @()
$cutoff = (Get-Date).AddDays(-60)
# 1. Find vendor/service accounts that are enabled but stale or never logged on
$vendorAccounts = Get-ADUser -Filter {
Enabled -eq $true -and (
Name -like '*vendor*' -or Name -like '*partner*' -or
Name -like '*svc-*' -or Name -like 'ext-*' -or Name -like '*thirdparty*')
} -Properties LastLogonDate, PasswordLastSet, PasswordNeverExpires, MemberOf, logonHours
foreach ($acct in $vendorAccounts) {
$stale = ($null -eq $acct.LastLogonDate) -or ($acct.LastLogonDate -lt $cutoff)
$report += [PSCustomObject]@{
SamAccountName = $acct.SamAccountName
LastLogon = $acct.LastLogonDate
PasswordAge = if ($acct.PasswordLastSet) { ((Get-Date) - $acct.PasswordLastSet).Days } else { 'Never set' }
PasswordNeverExpires = $acct.PasswordNeverExpires
IsStale = $stale
PrivilegedGroups = ($acct.MemberOf | Where-Object { $_ -match 'Admin|Remote Desktop' }) -join '; '
LogonWorkstations = $acct.logonHours -ne $null
}
}
$report | Sort-Object IsStale -Descending | Format-Table -AutoSize
$report | Export-Csv -Path ".\vendor_account_audit_$(Get-Date -Format 'yyyyMMdd').csv" -NoTypeInformation
# 2. Flag and optionally disable stale vendor accounts (uncomment after review)
# $report | Where-Object { $_.IsStale -and -not $_.PrivilegedGroups } | ForEach-Object {
# Disable-ADAccount -Identity $_.SamAccountName -WhatIf
# }
# 3. Enforce workst## Remediationation logon restrictions on one account (repeat per account — vendor accounts should
# only authenticate to designated jump/integration hosts)
# Set-ADUser -Identity 'svc-vendor-x' -LogonWorkstations 'VENDOR-JUMP01,INTEGRATION-SRV02'
# 4. Reset credentials for any account with no recorded logon restriction
# Assume compromise if the account was reachable from the vendor network during the window
# Set-ADAccountPassword -Identity 'svc-vendor-x' -Reset -NewPassword (Read-Host -AsSecureString 'NewPwd')
Write-Host "`n[+] Audit complete. Review the CSV, disable stale accounts, restrict logon scope," `
-ForegroundColor Green
Write-Host " and rotate credentials for any vendor account without workstation restrictions."
Remediation
There is no patch for a trusted-relationship breach — remediation is architectural and procedural. Prioritize the following:
Immediate (0–72 hours):
- Inventory every third-party identity and integration path. You cannot defend access you have not cataloged. Enumerate vendor accounts, API keys, service principals, SFTP endpoints, EDI pipelines, and support VPNs. The audit script above is a starting point.
- Rotate credentials for all third-party accounts, especially any without MFA or with unrestricted logon scope. If a vendor in your ecosystem is named in an extortion claim, assume credential compromise until proven otherwise.
- Disable stale vendor accounts (no logon in 60 days) pending owner re-validation.
- Enable and review verbose logging on vendor-accessible systems: authentication, file access on data repositories, and egress volume. If you cannot see what a vendor account touched, you cannot scope a breach.
Short term (1–2 weeks): 5. Enforce least privilege on third-party access. Vendor accounts should reach exactly the systems and data their contract requires — via workstation logon restrictions, firewall-scoped network segments, and just-in-time elevation. Standing broad access is how one vendor breach becomes your breach. 6. Deploy or tune the detections above. Baseline vendor account behavior (source IPs, hours, accessed systems, transfer volumes) and alert on deviation. Behavioral anomaly detection on trusted accounts is the single highest-value control for this threat class. 7. Monitor egress. Large outbound transfers from data-bearing servers to novel destinations should page a human. Extortion actors depend on quiet exfiltration — deny them that. 8. Watch the actor's leak infrastructure and credential markets for your domains, vendor names, and customer data signatures. If you are in the financial or payments ecosystem, treat N0n's claims as a direct prompt to run the hunt queries in this post.
Strategic (this quarter): 9. Rebuild third-party risk management around continuous verification, not annual questionnaires. Contractual attestations did not stop this breach. Require vendors to meet enforceable technical controls — MFA, logging, network segmentation, incident notification SLAs — and verify them. 10. Run a tabletop exercise on exactly this scenario: a named vendor reports compromise, and you have 4 hours to determine what data they could reach, what was touched, and what to revoke. If the answer takes longer than the exercise window, that gap is your remediation plan. 11. Review your breach-notification and regulatory obligations (GLBA, state privacy laws, PCI-DSS if cardholder data is in scope, and sector-specific regulators). JPMorgan's confirmation cycle is a preview of the disclosure clock you will be on.
For organizations in the payments ecosystem specifically: the claimed targeting of JPMorgan, Visa, and Mastercard suggests deliberate focus on interconnected financial infrastructure. Shared vendors — processors, integrators, support providers — are force multipliers for the actor and single points of failure for you. Map your vendor overlap with these institutions now.
Related Resources
Security Arsenal Incident Response Services AlertMonitor Platform Book a SOC Assessment incident-response Intel Hub
Is your security operations ready?
Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.