Levi Strauss & Co. has disclosed a cyberattack in which a threat actor used social engineering to gain access to the computers of three employees and exfiltrate corporate data from those systems. The incident, reported via SecurityWeek, is a textbook example of a trend that has defined the intrusion landscape through 2025 and into 2026: attackers are no longer leading with exploits — they're leading with phone calls.
There is no CVE here, no zero-day, no exotic implant. The vulnerability exploited was the human layer — likely the helpdesk, the employee, or both — and the blast radius was determined entirely by what those three accounts and endpoints could reach. That is precisely what makes this story worth a defender's full attention. When Scattered Spider–affiliated actors demonstrated in 2023–2025 that a well-placed phone call could net an entire enterprise identity stack, the lesson should have stuck: identity is the perimeter, and the service desk is now a Tier-1 attack surface.
This post breaks down the attack chain implied by the Levi Strauss disclosure, maps it to observable behaviors, and delivers detection content and hardening steps your SOC can implement this week.
Technical Analysis
What We Know
Per the disclosure, the attack chain was:
- Social engineering as initial access (TA0001 / T1566, T1078). The threat actor manipulated employees — directly or via internal support personnel — into granting access to their machines. In 2025–2026, this typically means one or more of: voice phishing (vishing) to the helpdesk to reset MFA or credentials, smishing/malicious links delivering remote access tooling, or MFA fatigue/prompt bombing until a user approves.
- Interactive access to three employee endpoints. Access to "the computers" implies either valid-account logon with stolen credentials, or remote access tooling (RATs such as AnyDesk, ScreenConnect, TeamViewer, or Atera) installed under the guise of IT support — a hallmark of helpdesk impersonation tradecraft.
- Data exfiltration (TA0010 / T1041, T1567). Corporate data was taken from those endpoints. In modern intrusions this is typically staged to cloud storage (MEGA, Dropbox, Google Drive), pushed over HTTPS to attacker infrastructure, or pulled directly from SaaS tenants using the compromised session.
Affected Surface
The affected "products" here are not vendor software — they are the identity and endpoint controls of the victim organization:
- Identity provider / MFA enrollment flow — the helpdesk-driven reset path is the weak link.
- Employee endpoints — three workstations with access to corporate data.
- Data egress paths — unsanctioned cloud storage, unmonitored HTTPS upload volume.
Exploitation Status
This is confirmed, real-world exploitation of a process, not a product. No CVE applies, and none should be invented. The technique cluster — helpdesk social engineering, RMM tool abuse, cloud exfiltration — is actively used by financially motivated threat actors tracked across 2025–2026 reporting, and maps to MITRE ATT&CK techniques T1078 (Valid Accounts), T1219 (Remote Access Software), T1078.004 (Cloud Accounts), T1567.002 (Exfiltration to Cloud Storage), and T1020/T1041 (Exfiltration Over C2/Alternative Channels).
Why Three Endpoints Is Not "Small"
Defenders should resist the temptation to read "three employees" as "minor." Three well-chosen accounts — an executive assistant, a finance user, an IT admin — can expose the entire crown-jewel data set. The actor picked those three for a reason, or pivoted to the highest-value targets their social engineering reached.
Detection & Response
The most reliable detections for this intrusion pattern do not hunt the phone call — they hunt the consequences: anomalous sign-ins, remote access tooling appearing on endpoints, and bulk egress behavior. The rules below target those observables.
Sigma Rules
The following rules detect (1) remote access tool installation/execution consistent with helpdesk impersonation tradecraft and (2) browser or process access to known personal/exfiltration cloud storage endpoints from corporate systems.
---
title: Remote Access Tool Execution - Helpdesk Impersonation Pattern
id: 3f8a1b2c-7d4e-4f91-a6c3-9e2b5d8f1a07
status: experimental
description: Detects execution of commonly abused remote access tools (AnyDesk, ScreenConnect, TeamViewer, Atera, Splashtop) frequently deployed via social engineering and helpdesk impersonation. Tune against an approved-RMM allowlist.
references:
- https://attack.mitre.org/techniques/T1219/
- https://www.securityweek.com/corporate-data-stolen-in-levi-strauss-cyberattack/
author: Security Arsenal
date: 2026/04/06
tags:
- attack.command_and_control
- attack.t1219
logsource:
category: process_creation
product: windows
detection:
selection_img:
Image|endswith:
- '\anydesk.exe'
- '\screenconnect.client.exe'
- '\teamviewer.exe'
- '\teamviewer_service.exe'
- '\atera_agent.exe'
- '\splashtop.exe'
- '\sr_manager.exe'
selection_cli:
CommandLine|contains:
- 'anydesk'
- 'screenconnect'
- 'teamviewer'
- 'atera'
- 'splashtop'
condition: 1 of selection_*
falsepositives:
- Approved RMM platforms deployed by internal IT - allowlist sanctioned tools and installation paths
level: high
---
title: Suspicious Cloud Storage Access for Data Exfiltration
id: 8c2e4d71-1a9b-4c55-b3e8-6f0d2a7c4b19
status: experimental
description: Detects processes associated with file staging or browser access reaching consumer cloud storage services commonly abused for data exfiltration (MEGA, Dropbox, WeTransfer, file.io). Correlate with upload volume.
references:
- https://attack.mitre.org/techniques/T1567/002/
- https://www.securityweek.com/corporate-data-stolen-in-levi-strauss-cyberattack/
author: Security Arsenal
date: 2026/04/06
tags:
- attack.exfiltration
- attack.t1567.002
logsource:
category: network_connection
product: windows
detection:
selection:
DestinationHostname|contains:
- '.mega.nz'
- '.mega.io'
- '.dropbox.com'
- '.wetransfer.com'
- '.file.io'
- '.transfer.sh'
- '.gofile.io'
- '.pixeldrain.com'
condition: selection
falsepositives:
- Business-sanctioned use of file transfer services - restrict to unsanctioned destinations via proxy policy and allowlist approved tenants
level: medium
---
title: Archive Utility Execution Followed by Browser Activity - Staging Pattern
id: 5b1d9e43-8f2a-4c76-9d31-4e8a6c0b7f52
status: experimental
description: Detects execution of archive utilities (rar, 7z, winrar) with compression arguments, a common data-staging step before exfiltration of collected corporate files from endpoints.
references:
- https://attack.mitre.org/techniques/T1560/001/
- https://www.securityweek.com/corporate-data-stolen-in-levi-strauss-cyberattack/
author: Security Arsenal
date: 2026/04/06
tags:
- attack.collection
- attack.t1560.001
logsource:
category: process_creation
product: windows
detection:
selection:
Image|endswith:
- '\rar.exe'
- '\7z.exe'
- '\7za.exe'
- '\winrar.exe'
CommandLine|contains:
- ' a '
- ' -r'
- ' -p'
condition: selection
falsepositives:
- Legitimate backup and software packaging workflows - baseline per-user archive tool usage and alert on first-seen
level: medium
KQL — Microsoft Sentinel / Defender
This hunt correlates the three legs of the intrusion: sign-in anomalies, RMM execution, and egress to unsanctioned storage. Run each section during an investigation; join on account and device to build the timeline.
// Hunt 1: First-seen remote access tool execution per device (last 14 days vs 30-day baseline)
let lookback = 14d;
let baseline = 30d;
let rmm_tools = dynamic(["anydesk.exe","screenconnect.client.exe","teamviewer.exe","atera_agent.exe","splashtop.exe","sr_manager.exe"]);
let known = DeviceProcessEvents
| where TimeGenerated between (ago(baseline) .. ago(lookback))
| where FileName in~ (rmm_tools)
| summarize by DeviceName, FileName;
DeviceProcessEvents
| where TimeGenerated > ago(lookback)
| where FileName in~ (rmm_tools)
| where not(known has (DeviceName, FileName))
| project TimeGenerated, DeviceName, AccountName, FileName, ProcessCommandLine, InitiatingProcessAccountName, InitiatingProcessFileName
| sort by TimeGenerated desc;
// Hunt 2: High-volume outbound transfers to unsanctioned cloud storage (exfil staging)
DeviceNetworkEvents
| where TimeGenerated > ago(7d)
| where RemoteUrl has_any ("mega.nz","mega.io","wetransfer.com","file.io","gofile.io","transfer.sh","pixeldrain.com")
| summarize ConnectionCount = count(), FirstSeen = min(TimeGenerated), LastSeen = max(TimeGenerated)
by DeviceName, InitiatingProcessFileName, RemoteUrl, RemoteIP
| where ConnectionCount > 20
| sort by ConnectionCount desc;
// Hunt 3: MFA reset or auth-method change followed by new sign-in from unusual location (helpdesk social engineering indicator)
AuditLogs
| where TimeGenerated > ago(14d)
| where OperationName has_any ("Update user","Reset user password","Update authentication methods")
| extend TargetUser = tostring(TargetResources[0].userPrincipalName)
| extend Actor = tostring(InitiatedBy.user.userPrincipalName)
| project ResetTime = TimeGenerated, Actor, TargetUser, OperationName
| join kind=inner (
SigninLogs
| where TimeGenerated > ago(14d)
| where ResultType == 0
| project SigninTime = TimeGenerated, UserPrincipalName, IPAddress, LocationDetails, AppDisplayName
) on $left.TargetUser == $right.UserPrincipalName
| where SigninTime between (ResetTime .. ResetTime + 2h)
| project ResetTime, Actor, TargetUser, OperationName, SigninTime, IPAddress, LocationDetails, AppDisplayName
| sort by ResetTime desc;
Velociraptor VQL
Use this artifact to sweep the fleet for RMM binaries in non-standard paths and recently created archives — the two on-disk artifacts most likely to survive after a social-engineering-driven intrusion.
-- Hunt: RMM tools outside sanctioned paths and recently created archives
LET rmm = SELECT FullPath, Size, Mtime
FROM glob(globs=[
'C:/Users/*/Downloads/**/anydesk*.exe',
'C:/Users/*/Downloads/**/teamviewer*.exe',
'C:/Users/*/Downloads/**/screenconnect*.exe',
'C:/Users/*/AppData/**/atera*.exe',
'C:/ProgramData/**/splashtop*.exe'
])
WHERE NOT FullPath =~ '(?i)Program Files'
LET archives = SELECT FullPath, Size, Mtime
FROM glob(globs=['C:/Users/*/**/*.zip','C:/Users/*/**/*.rar','C:/Users/*/**/*.7z'])
WHERE Mtime > now() - 604800
AND Size > 10485760
SELECT * FROM rmm
UNION ALL
SELECT * FROM archives
Remediation & Verification Script
Run this PowerShell on endpoints in scope (or deploy via your RMM/Intune) to enumerate RMM persistence, unsanctioned tools, and recently staged archives, then remove unauthorized services.
# Levi Strauss-pattern IR sweep: RMM artifacts, persistence, staged archives
# Run elevated. Review output before removing anything.
$report = @()
# 1. Enumerate services and auto-runs tied to common RMM tooling
$rmmNames = 'AnyDesk|ScreenConnect|TeamViewer|Atera|Splashtop|ConnectWise'
$services = Get-CimInstance Win32_Service | Where-Object { $_.Name -match $rmmNames -or $_.PathName -match $rmmNames }
foreach ($s in $services) {
$report += [pscustomobject]@{ Type='Service'; Name=$s.Name; Path=$s.PathName; State=$s.State; StartMode=$s.StartMode }
}
# 2. Check Run keys for RMM persistence
$runKeys = 'HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Run',
'HKCU:\SOFTWARE\Microsoft\Windows\CurrentVersion\Run',
'HKLM:\SOFTWARE\WOW6432Node\Microsoft\Windows\CurrentVersion\Run'
foreach ($key in $runKeys) {
if (Test-Path $key) {
Get-ItemProperty $key | Get-Member -MemberType NoteProperty | Where-Object { $_.Name -match $rmmNames } | ForEach-Object {
$report += [pscustomobject]@{ Type='RunKey'; Name=$_.Name; Path=(Get-ItemPropertyValue $key -Name $_.Name); State='Persistent'; StartMode='Auto' }
}
}
}
# 3. Find archives >10MB created in the last 7 days under user profiles (staging indicator)
$cutoff = (Get-Date).AddDays(-7)
Get-ChildItem 'C:\Users' -Recurse -Include *.zip,*.rar,*.7z -ErrorAction SilentlyContinue |
Where-Object { $_.LastWriteTime -gt $cutoff -and $_.Length -gt 10MB } | ForEach-Object {
$report += [pscustomobject]@{ Type='StagedArchive'; Name=$_.Name; Path=$_.FullName; State="$([math]::Round($_.Length/1MB,1)) MB"; StartMode=$_.LastWriteTime }
}
# 4. Output and optional removal of unapproved RMM services (uncomment to enforce)
$report | Format-Table -AutoSize
$report | Export-Csv ".\rmm_sweep_$(Get-Date -Format yyyyMMdd_HHmm).csv" -NoTypeInformation
# foreach ($s in $services) {
# Stop-Service -Name $s.Name -Force -ErrorAction SilentlyContinue
# Set-Service -Name $s.Name -StartupType Disabled
# sc.exe delete $s.Name
# }
Remediation
Because this attack exploited process and identity rather than software, remediation is architectural:
- Harden the helpdesk identity-verification workflow immediately. Require out-of-band verification for MFA resets and password resets on privileged or sensitive accounts: manager callback to a directory-listed number, hardware-token re-enrollment only in person or via verified video session, and mandatory ticketing with dual approval for MFA changes. Audit every reset performed in the last 90 days and cross-reference against sign-in telemetry.
- Deploy phishing-resistant MFA (FIDO2/passkeys) and number matching. Push-based MFA with prompt bombing remains the most common bypass. Number matching defeats fatigue attacks; FIDO2 defeats adversary-in-the-middle phishing outright. Prioritize helpdesk staff, admins, and finance roles.
- Allowlist remote access tooling. Block execution of all RMM binaries except your sanctioned platform via AppLocker/WDAC or your EDR's application control. An unsanctioned AnyDesk install should page the on-call analyst, not succeed.
- Constrain egress. Enforce proxy-based controls that block unsanctioned cloud storage destinations (MEGA, WeTransfer, file.io, etc.) and alert on upload volume anomalies per user/device. If you use a sanctioned file-transfer service, allowlist only the corporate tenant.
- Scope the data exposure forensically. For the three affected endpoints (and any analogous incident in your environment), build the timeline: what data resided locally, what SaaS sessions were active, what was uploaded and where. Notify per your regulatory obligations — for a company like Levi Strauss, that analysis spans employment, customer, and partner data regimes.
- Run targeted vishing exercises. Tabletop and live social-engineering drills against the helpdesk are the single highest-ROI control validation for this exact attack pattern. Measure: did the agent follow verification procedure, and did they report the attempt?
There is no patch to install and no CISA KEV deadline to meet — the deadline is your next helpdesk shift. Treat identity-reset procedures with the same change-control rigor you apply to production firewalls.
Related Resources
Security Arsenal Healthcare Cybersecurity AlertMonitor Platform Book a SOC Assessment healthcare Intel Hub
Is your security operations ready?
Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.