New reporting confirms that the threat actor tracked as Jewelbug has been conducting sustained espionage operations against government and military organizations — including a months-long compromise of a government webmail environment — while simultaneously operating cryptocurrency fraud schemes for financial gain. That combination should get your attention. When an intrusion set is motivated by both intelligence collection and monetization, the dwell time extends, the tooling diversifies, and the blast radius grows well beyond the initial victim.
Webmail servers are a high-value target precisely because defenders under-instrument them. They sit at the perimeter, they hold the crown jewels (executive communications, credentials, attachments, password-reset flows), and in many organizations they log far less than endpoints do. Jewelbug's campaign is a case study in why that gap is dangerous: long-term mailbox access enables silent intelligence collection, and the same foothold can be leveraged for financially motivated side operations — fraud, resale of access, or crypto-themed social engineering — without tripping the alarms a ransomware deployment would.
If you operate government, defense, or any high-value enterprise email infrastructure, assume this tradecraft is being replicated and act on the guidance below.
Technical Analysis
What Jewelbug Did
Based on the published reporting, Jewelbug's operations are characterized by:
- Extended dwell time in government email infrastructure. The group maintained access to a government webmail environment for months, indicating successful persistence, credential theft, and disciplined operational security — the hallmarks of a mature espionage actor rather than an opportunistic criminal crew.
- Target set: governments and militaries. This is intelligence-collection tradecraft. Mailbox contents, contact graphs, and credentials harvested from webmail provide leverage for downstream spearphishing and lateral movement into connected networks.
- Parallel cryptocurrency fraud. Alongside espionage, the group engaged in crypto fraud — a dual-motivation model we've seen from several state-adjacent and hybrid actors. Financially motivated activity can fund operations and also serves as cover noise that muddies attribution.
- Living-off-the-land and commodity tooling. Consistent with Jewelbug's previously documented operations, the actor favors legitimate system binaries, renamed tools, and tunneling mechanisms to blend C2 traffic into normal outbound web traffic — deliberately minimizing the forensic footprint on hosts that often lack EDR coverage (webmail servers, mail gateways).
Why Webmail Compromise Is the Centerpiece
A webmail breach is not a single event — it is a capability the attacker retains:
- Collection: Adversaries read mail in real time or silently sync mailbox contents (IMAP/OWA/EWS abuse) without deploying malware to endpoints.
- Credential harvesting: Password resets, VPN enrollment emails, and MFA setup flows all transit email. Control of mailboxes frequently equals control of adjacent identities.
- Trusted-channel phishing: Internal spearphishing from a legitimate government mailbox bypasses most user skepticism and many email gateways.
- Persistence via mail rules: Attackers create hidden inbox forwarding/deletion rules to exfiltrate specific threads and bury the evidence.
Exploitation Status
This is confirmed active exploitation in the wild — an observed, sustained intrusion against production government infrastructure, not a theoretical technique. No CVE has been publicly associated with this campaign in the reporting; initial access vectors for Jewelbug historically include internet-facing application exploitation and valid credential use. Defenders should treat unpatched, internet-facing mail and remote-access infrastructure as the presumed entry point and audit accordingly.
Detection & Response
The detections below target the behaviors that matter in this campaign: unauthorized mailbox access and rule manipulation, persistence on mail/web servers, credential theft from server hives, and tunneling-based C2. Tune thresholds to your environment — these are hunting-grade, not fire-and-forget.
Sigma Rules
---
title: Suspicious Inbox Rule Creation or Forwarding on Mail Server
description: Detects creation of inbox forwarding or deletion rules indicative of mailbox-based exfiltration, consistent with Jewelbug-style webmail compromise. Fires on Exchange/OWA rule manipulation via PowerShell remoting or EWS.
references:
- https://attack.mitre.org/techniques/T1114/002/
- https://attack.mitre.org/techniques/T1098/002/
author: Security Arsenal
date: 2026/04/06
status: experimental
tags:
- attack.collection
- attack.t1114.002
- attack.persistence
logsource:
product: windows
category: process_creation
detection:
selection_cmdlets:
CommandLine|contains:
- 'New-InboxRule'
- 'Set-InboxRule'
- 'Enable-InboxRule'
selection_actions:
CommandLine|contains:
- 'ForwardTo'
- 'ForwardAsAttachmentTo'
- 'RedirectTo'
- 'DeleteMessage'
condition: selection_cmdlets and selection_actions
falsepositives:
- Legitimate helpdesk mailbox administration; correlate with change tickets and admin account baselines
level: high
---
title: SAM or System Hive Dump on Mail or Web Server
description: Detects credential theft via registry hive export (reg save of SAM/SYSTEM/SECURITY) on servers hosting webmail or web applications, a technique used to extract local credentials for persistence and lateral movement.
references:
- https://attack.mitre.org/techniques/T1003/002/
author: Security Arsenal
date: 2026/04/06
status: experimental
tags:
- attack.credential_access
- attack.t1003.002
logsource:
product: windows
category: process_creation
detection:
selection_img:
Image|endswith: '\reg.exe'
selection_cli:
CommandLine|contains:
- ' save '
- ' export '
selection_hives:
CommandLine|contains:
- 'HKLM\SAM'
- 'HKLM\SYSTEM'
- 'HKLM\SECURITY'
- 'hkey_local_machine\sam'
- 'hkey_local_machine\system'
- 'hkey_local_machine\security'
condition: selection_img and selection_cli and selection_hives
falsepositives:
- Rare legitimate backup or migration activity; validate against authorized admin windows
level: critical
---
title: Tunneling or Reverse Proxy Tool Execution on Server
description: Detects execution of common tunneling/ingress tools (cloudflared, ngrok, chisel, frpc, ligolo) frequently abused by APT actors including Jewelbug for covert C2 channels that blend into outbound HTTPS.
references:
- https://attack.mitre.org/techniques/T1572/
- https://attack.mitre.org/techniques/T1090/
author: Security Arsenal
date: 2026/04/06
status: experimental
tags:
- attack.command_and_control
- attack.t1572
- attack.t1090
logsource:
category: process_creation
detection:
selection_name:
Image|endswith:
- '\cloudflared.exe'
- '\ngrok.exe'
- '\chisel.exe'
- '\frpc.exe'
- '\ligolo-agent.exe'
- '\ligolo-ng.exe'
- '/cloudflared'
- '/ngrok'
- '/chisel'
- '/frpc'
selection_cli:
CommandLine|contains:
- ' tunnel '
- 'trycloudflare.com'
- ' --url '
- 'socks'
- 'reverse'
condition: selection_name or selection_cli
falsepositives:
- Legitimate developer or DevOps tunneling; maintain an approved-tools inventory and alert on deviation
level: high
KQL — Microsoft Sentinel / Defender
Hunt for anomalous external access to mailboxes, inbox rule creation, and suspicious child processes under mail/web server processes. The first query targets mailbox access anomalies via Office 365 audit data; the second hunts server-side execution on Linux/Windows mail infrastructure ingested via Syslog/CEF.
// Hunt 1: Suspicious inbox rule creation and mailbox forwarding (Exchange Online / O365 audit)
OfficeActivity
| where TimeGenerated > ago(30d)
| where Operation in ("New-InboxRule", "Set-InboxRule", "Set-Mailbox", "Add-MailboxPermission")
| extend Parameters = tostring(OfficeObjectId)
| extend Params = parse_json(tostring(parse_json(RecordType)))
| mv-expand parse_json(parse_json(tostring(ParametersParsed = "") )) // placeholder-safe: use raw JSON below
| extend RawParams = tostring(parse_json(tostring("")))
| project TimeGenerated, UserId, ClientIP, Operation, OfficeObjectId, ResultStatus
| where ResultStatus =~ "Success"
| summarize RuleChanges = count(), Operations = make_set(Operation), SourceIPs = make_set(ClientIP) by UserId, bin(TimeGenerated, 1d)
| where RuleChanges > 2
| order by TimeGenerated desc;
// Hunt 2: Non-standard client IPs accessing many distinct mailboxes (IMAP/OWA/EWS abuse)
OfficeActivity
| where TimeGenerated > ago(14d)
| where Operation in ("MailItemsAccessed", "UserLoggedIn", "MailboxLogin")
| summarize DistinctMailboxes = dcount(OfficeObjectId), IPs = make_set(ClientIP) by UserId, bin(TimeGenerated, 1h)
| where DistinctMailboxes > 10
| order by DistinctMailboxes desc;
// Hunt 3: Web server processes spawning shells or credential tools (Linux mail/web servers via Syslog)
Syslog
| where TimeGenerated > ago(14d)
| where ProcessName in ("reg", "cmd", "powershell", "pwsh", "bash", "sh", "openssl", "cloudflared", "ngrok", "chisel")
| where SyslogMessage has_any ("save HKLM", "shadow", "htpasswd", "tunnel", "reverse", "--url")
| project TimeGenerated, Computer, ProcessName, SyslogMessage
| order by TimeGenerated desc;
Note: Hunt 1 uses the OfficeActivity table; if your tenant ingests Exchange admin audit logs into a custom table, adapt the Operation values accordingly. Validate parsing of the Parameters field in your workspace — rule-creation parameters (ForwardTo addresses) are the highest-fidelity artifact.
Velociraptor VQL
Hunt mail and web servers for tunneling tools, hive dumps staged on disk, and suspicious outbound connections.
-- Hunt: tunneling tools, credential dump artifacts, and anomalous connections on servers
LET procs = SELECT Pid, Name, CommandLine, Exe, Username, CreateTime
FROM pslist()
WHERE CommandLine =~ '(?i)(tunnel|trycloudflare|reverse|socks|--url|reg save|save HKLM)'
OR Exe =~ '(?i)(cloudflared|ngrok|chisel|frpc|ligolo)'
LET artifacts = SELECT FullPath, Size, Mtime
FROM glob(globs=['C:/Windows/Temp/*.hiv', 'C:/Windows/Temp/*.sam', 'C:/Windows/Temp/*.save',
'C:/ProgramData/**/*.hiv', '/tmp/*.hiv', '/tmp/sam*', '/var/tmp/*.save'])
LET conns = SELECT Pid, Name, Status, RaddrIP = Raddr.IP, RaddrPort = Raddr.Port
FROM netstat()
WHERE Status =~ 'ESTABLISHED'
AND RaddrPort in (443, 7844, 2053, 2083, 2087, 2096)
AND NOT Name =~ '(?i)(w3wp|nginx|apache|httpd|store|edge|chrome)'
SELECT * FROM procs
UNION ALL SELECT NULL AS Pid, 'FILE_ARTIFACT' AS Name, FullPath AS CommandLine,
str(Size) AS Exe, '' AS Username, Mtime AS CreateTime FROM artifacts
Remediation & Verification Script
Run this PowerShell audit on Exchange/mail servers and domain controllers to surface Jewelbug-style artifacts: forwarding rules, rogue admins, tunneling tools, and hive-dump staging files.
# Jewelbug-Style Intrusion Audit — Exchange/Mail Servers
# Run elevated. Review output before taking action.
$report = @{}
# 1. Audit inbox forwarding rules across all mailboxes (Exchange on-prem/Online)
try {
$suspiciousRules = Get-Mailbox -ResultSize Unlimited | ForEach-Object {
Get-InboxRule -Mailbox $_.Alias -ErrorAction SilentlyContinue |
Where-Object { $_.ForwardTo -or $_.ForwardAsAttachmentTo -or $_.RedirectTo -or $_.DeleteMessage }
}
$report['SuspiciousInboxRules'] = $suspiciousRules | Select-Object MailboxOwnerId, Name, ForwardTo, RedirectTo, DeleteMessage
} catch { Write-Warning "Exchange cmdlets unavailable: $_" }
# 2. Check for tunneling tools running on the host
$report['TunnelingProcesses'] = Get-Process -ErrorAction SilentlyContinue |
Where-Object { $_.Name -match 'cloudflared|ngrok|chisel|frpc|ligolo' } |
Select-Object Name, Id, Path, StartTime
# 3. Look for staged credential hive dumps in common staging dirs
$report['PossibleHiveDumps'] = Get-ChildItem -Path "$env:TEMP","C:\Windows\Temp","C:\ProgramData" -Recurse -ErrorAction SilentlyContinue -Depth 2 |
Where-Object { $_.Extension -in '.hiv','.sam','.save' } |
Select-Object FullName, Length, LastWriteTime
# 4. Enumerate recently created local admins (persistence check)
$report['LocalAdmins'] = Get-LocalGroupMember -Group 'Administrators' -ErrorAction SilentlyContinue |
Select-Object Name, ObjectClass, PrincipalSource
# 5. Recently created scheduled tasks (persistence)
$report['RecentScheduledTasks'] = Get-ScheduledTask | Where-Object {
$_.Date -and ([datetime]$_.Date) -gt (Get-Date).AddDays(-90)
} | Select-Object TaskName, TaskPath, Date
# 6. Unauthorized OpenSSH server installation
$ssh = Get-WindowsCapability -Online -Name 'OpenSSH.Server*' -ErrorAction SilentlyContinue
$report['OpenSSHServerState'] = $ssh.State
# Export findings
$report.GetEnumerator() | ForEach-Object {
Write-Host "`n=== $($_.Key) ===" -ForegroundColor Cyan
$_.Value | Format-List
}
$report | ConvertTo-Json -Depth 4 | Out-File "$env:USERPROFILE\Desktop\jewelbug_audit_$(Get-Date -Format yyyyMMdd_HHmm).json"
Write-Host "`nAudit complete. JSON saved to Desktop." -ForegroundColor Green
Remediation
If you find evidence consistent with this activity — or want to preempt it — execute in this order:
- Assume credential compromise and rotate aggressively. Reset passwords for all mail-enabled accounts, prioritizing administrators and service accounts. Invalidate active sessions and OAuth tokens (Exchange Online:
Revoke-AzureADUserAllRefreshToken/RevokeSignInSessions). Jewelbug's dwell time means credentials harvested months ago may still be valid. - Audit and purge inbox rules tenant-wide. Malicious forwarding rules survive password resets. Export all rules, flag external forwarders and auto-delete rules, and disable rules created outside approved change windows.
- Lock down external email access protocols. Disable legacy IMAP/POP/basic auth where not strictly required. Enforce MFA on all webmail/OWA access and apply conditional access to block logins from unmanaged networks and anonymizing infrastructure.
- Patch all internet-facing mail and remote-access infrastructure. Jewelbug's historical initial-access tradecraft favors exposed applications and valid accounts. Inventory every internet-facing service, confirm current patch levels, and remove anything that doesn't need to be reachable from the internet. Monitor CISA KEV weekly for newly exploited mail-edge and VPN products.
- Deploy EDR and logging on mail servers — not just endpoints. Enable command-line process auditing (Event 4688 / Sysmon), forward Exchange admin audit logs and IIS/OWA logs to your SIEM, and alert on non-admin mailbox access at scale.
- Hunt for tunnels and persistence. Block unauthorized tunneling tools via application control (WDAC/AppLocker), alert on outbound connections to dynamic tunnel domains, and review scheduled tasks, services, and local admin group membership on all servers.
- Brief finance and executives on the crypto-fraud angle. Dual-motivation actors weaponize trusted mailboxes for business email compromise and crypto-themed fraud. Enforce out-of-band verification for payment and wallet-address changes — full stop.
- Engage IR if indicators confirm. A months-long dwell time means eradication requires a formal scoping exercise: full credential reset, persistence sweep, log retention review, and monitoring of re-entry attempts for at least 90 days post-eviction.
Related Resources
Security Arsenal Red Team Services AlertMonitor Platform Book a SOC Assessment pen-testing Intel Hub
Is your security operations ready?
Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.