Attack scenario · Kubernetes & infrastructure

Resource hijacking and unauthorized compute use

How do attackers abuse AWS accounts for cryptomining or AI compute? Stolen credentials are used to launch large GPU or compute fleets — often in regions the company never uses — leaving the victim with the bill.

Fictional, educational scenarioFinancial abuseService disruptionLast updated · ATT&CK v20.1

Attack path · 3 stages

  1. 1

    Use stolen credentials

    T1078.004

  2. 2

    Move to unused regions

    T1535

  3. 3

    Launch compute for the adversary’s use

    T1496

For leadership

Executive summary

Not every attacker wants your data. Many simply want your compute — to mine cryptocurrency, run their own workloads, or consume paid AI model capacity — and you pay for it. Because these attacks often run in regions you never use, they may only be noticed when the bill arrives. Simple measures — denying unused regions, keeping service quotas realistic and alerting on unusual spend — limit both the likelihood and the financial impact. Resource hijacking should also be treated as evidence of a wider credential compromise: whoever launched the instances held working credentials and may have used them for other purposes too.

For engineers

Technical summary

With ec2:RunInstances (or Lambda, ECS, Bedrock invoke permissions) an adversary launches resources for their own use (T1496). Unused regions (T1535) and quota increase requests are common accompaniments.

Environment and assumptions

Fictional company: a leaked developer access key has ec2:* in a sandbox account with default service quotas. No region restrictions exist.

  • The adversary has a key with compute-launch permissions.
  • Fictional and simplified for education.

Interactive

Attack sequence

Select a stage, or use the arrow keys, to see preconditions, evidence and controls.

Stage 1

Use stolen credentials

The key is validated and permissions probed.

Preconditions
  • Active key
Required access
  • Key.
Potential evidence
  • CloudTrail from unfamiliar infrastructure
Defensive controls
  • Short-lived credentials
Text version of the attack sequence
  1. Use stolen credentials (T1078.004). The key is validated and permissions probed. Preconditions: Active key. Evidence: CloudTrail from unfamiliar infrastructure. Controls: Short-lived credentials.
  2. Move to unused regions (T1535). Resources are launched in regions the team does not monitor. Preconditions: No region-restricting SCP. Evidence: CloudTrail events in unused regions (multi-region trail required). Controls: SCP denying unused regions; GuardDuty enabled in all regions.
  3. Launch compute for the adversary’s use (T1496). Large or GPU instances are launched; quota increases may be requested. Preconditions: ec2:RunInstances; sufficient quota. Evidence: CloudTrail: RunInstances bursts; servicequotas:RequestServiceQuotaIncrease; Cost anomaly alerts; GuardDuty cryptocurrency findings. Controls: Instance-type allow-lists; Realistic quotas; Cost anomaly detection.

Technique mapping

MITRE ATT&CK and ATLAS techniques referenced in this scenario
IDTechniqueTactics (official)Stages
T1078.004Valid Accounts: Cloud AccountsInitial Access, Persistence, Privilege Escalation, Stealth1
T1535Unused/Unsupported Cloud RegionsStealth2
T1496Resource HijackingImpact3

Technique names and tactic assignments come from the imported MITRE data. Mapping a stage to a technique is Securitain’s analysis.

Why the attack can succeed

  • Unused regions are unmonitored.
  • Default quotas allow substantial spend.
  • Billing alerts are monthly rather than near-real-time.

What prevents or limits the attack

What may detect the attack

Detection signals and telemetry sources
SignalTelemetry sourceNotes
Compute launches in unused regionsMulti-region CloudTrail—
Spend anomaliesCost Anomaly Detection—
Mining-related network activityGuardDuty—

Telemetry only helps if it is enabled before the incident and someone acts on it. Listing a signal here does not mean it is collected or monitored in your environment.

What Securitain may assess

  • Securitain can assess todayCredential prioritization — surfacing access keys and identities whose permissions make a leak more damaging. Learn more
  • Securitain can assess todayEffective-permission view that considers managed, inline, group, boundary and SCP policies together. Learn more
  • Outside Securitain’s current scopeCost anomaly monitoring and compute runtime detection are not part of Securitain’s current assessment.

For leadership

Executive implications

operations
Unexpected cloud spend and potential quota exhaustion for legitimate workloads.
availability
Quota exhaustion can block legitimate scaling.

No financial loss estimates are given; impact depends on the data, contracts and regulation involved.

Next steps

  1. Apply a region-deny SCP.
  2. Enable cost anomaly alerts.
  3. Enable GuardDuty in every region.

References

MITRE ATT&CK® and ATT&CK® are registered trademarks of The MITRE Corporation. ATT&CK data © 2026 The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation (ATT&CK v20.1). Securitain is not affiliated with or endorsed by MITRE. Explanations, scenarios and control mappings on this page are Securitain’s own analysis. Sources and licensing

Stay in the loop

Cloud security insights, early access & compliance updates

Weekly InsightsEarly AccessExpert Content
No spam, everUnsubscribe anytimeWeekly updates
Chat with Securitain