MITRE ATT&CK technique · T1078.004

Valid Accounts: Cloud Accounts

How are valid cloud accounts abused in AWS attacks? Adversaries use legitimate cloud identities — stolen access keys, phished federated sessions, abused roles — so that their activity is authorized by design. Because every API call is technically permitted, defenders must rely on context (who, from where, doing what) rather than on blocked actions.

Initial AccessPersistencePrivilege EscalationStealthPlatforms: IaaS, Identity Provider, Office Suite, SaaSATT&CK v20.1 · source modified · View on MITRE ATT&CK

For leadership

Executive takeaway

Most cloud intrusions use real credentials. Reducing long-lived credentials, enforcing MFA, and knowing what each identity can reach matter more than any single detection.

How it works in AWS

  1. A credential is obtained through phishing, leaks in code or images, malware on a developer device, or a compromised third party.
  2. The adversary validates it (sts:GetCallerIdentity) and explores what it can access.
  3. Because the identity is legitimate, the activity is logged as normal API use; only context reveals it.

Sub-technique of T1078 Valid Accounts.

What the attacker needs

  • A usable credential or session
  • No network or device conditions blocking its use

Relevant AWS API calls

AWS API calls relevant to T1078.004
API / eventWhy it matters
sts:GetCallerIdentityValidate a credential; requires no permissions.
sts:AssumeRoleMove from one identity to another through trust.
ConsoleLogin (CloudTrail event)Console sign-ins, including whether MFA was used.

A risky permission is not an observed attack

Potential exposure

Active access keys older than policy, users without MFA, identities with broad permissions and no recent use.

Observed behavior

A credential used from a new ASN or country, GetCallerIdentity followed by enumeration bursts, sign-in without expected MFA context.

Preventive and limiting controls

  • Phishing-resistant MFA for humans
  • Replace IAM user keys with roles and federation
  • Retire unused identities
  • Restrict credential use by network where feasible (aws:SourceIp, aws:SourceVpce)

Mapped cloud controls

  • preventivePhishing-resistant MFA for human privileged access

    Stolen passwords alone cannot complete an interactive sign-in when MFA is enforced for the account.

  • restrictiveReplace long-term access keys with temporary credentials

    Short-lived credentials expire, so a leaked credential is useful for hours rather than indefinitely.

  • restrictiveFind and remove unused credentials and permissions

    Dormant users and keys are attractive because nobody notices when they start being used.

  • detectiveAlerting on identity and logging changes

    Alerts on unusual sign-ins, root use and new credentials can surface account misuse early.

  • detectiveManaged threat detection across accounts and regions

    GuardDuty can flag credential use from unusual locations or anonymizing infrastructure.

See assumptions and validation evidence in the Cloud Control Explorer →

Detection signals and telemetry

  • First use of an identity from a new network or user agent — CloudTrail; GuardDuty
  • Root user activity — CloudTrail
  • Dormant identity suddenly active — IAM last-used data; CloudTrail

MITRE detection strategies (official, cloud-relevant analytics)

  • DET0546 Detection of Abused or Compromised Cloud Accounts for Access and Persistence
    • AN1503 (Identity Provider): Detects anomalous authentication activity such as sign-ins from impossible geolocations or legacy protocols from high-privileged accounts.Log sources: azure:signinlogs — Sign-in activity; saas:okta — user.authentication.sso
    • AN1504 (IaaS): Detects cloud account use for API calls that exceed normal scope, such as IAM changes or access to services never used before.Log sources: AWS:CloudTrail — ConsoleLogin, AssumeRole, ListAccessKeys, CreateUser; gcp:audit — admin.googleapis.com
    • AN1505 (SaaS): Detects unexpected access or usage of cloud productivity tools (e.g., downloading large numbers of files, creating external shares) by internal users.Log sources: m365:unified — FileAccessed, SharingSet; gcp:audit — drive.activity
    • AN1506 (Office Suite): Detects login and usage patterns deviating from typical Microsoft 365 or Google Workspace user profiles.Log sources: m365:signinlogs — UserLogin; gcp:audit — login.event

MITRE mitigations

Official MITRE description (excerpt)

Valid accounts in cloud environments may allow adversaries to perform actions to achieve Initial Access, Persistence, Privilege Escalation, or Defense Evasion. Cloud accounts are those created and configured by an organization for use by users, remote support, services, or for administration of resources within a cloud service provider or SaaS application. Cloud Accounts can exist solely in the cloud; alternatively, they may be hybrid-joined between on-premises systems and the cloud through syncing or federation with other identity sources such as Windows Active Directory.

Common misconceptions

“Logged activity from a valid user is legitimate.”
Authorization is not intent. Context and baselines are required.

AI security

Related MITRE ATLAS techniques

ATLAS and ATT&CK are separate frameworks. These relationships are listed explicitly, with their basis.

Agent identity, permissions and credential exposure →Model and AI service access management →

Attack scenarios using this technique

Compromised AWS access key

A long-lived IAM user access key leaks from a developer laptop or repository and is used to explore the account, add persistence and read data.
Scenario

Dangerous cross-account role trust

A role trusts an entire external account without conditions. When that account is compromised, the adversary walks straight into yours.
Scenario

Cloud credential persistence

After initial access, the adversary creates additional keys, login profiles and roles so that rotating the original credential does not end the intrusion.
Scenario

Privilege escalation through IAM policy relationships

A developer role that can pass roles to Lambda and create functions can run code as a far more privileged role — without ever being granted administrator rights directly.
Scenario

Compromised deployment pipeline role

A CI/CD system that deploys to production with a broad role becomes a path from a code repository to production infrastructure.
Scenario

Exposed machine identity credentials

Service credentials embedded in a container image, environment variable or log file are harvested and reused from outside the organization.
Scenario

AWS IAM Identity Center permission misuse

An identity able to edit permission sets or account assignments grants itself administrator access across many accounts at once.
Scenario

Cross-account data movement

Rather than downloading data, an adversary shares snapshots or replicates buckets to an account they control — fast, quiet and inside AWS.
Scenario

CloudTrail logging impairment

Before acting, an adversary with administrative rights stops or narrows CloudTrail logging so that later activity leaves little evidence.
Scenario

Compromised EKS service account

A pod’s service account token is stolen; through Kubernetes RBAC and the IAM role mapped to it, the adversary reaches both the cluster and AWS.
Scenario

Overprivileged infrastructure automation

Terraform or CloudFormation automation runs with administrator rights across every account; whoever controls its inputs controls the organization.
Scenario

Resource hijacking and unauthorized compute use

Stolen credentials are used to launch large GPU or compute fleets — often in regions the company never uses — leaving the victim with the bill.
Scenario

AI agent with excessive AWS permissions

An operations agent that can call AWS APIs runs under a broad role. Manipulated input causes it to make changes no human approved.
Scenario

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). MITRE ATLAS™ data (v2026.10) © MITRE, used under the Apache License 2.0. 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