MITRE ATT&CK technique · T1098.003

Account Manipulation: Additional Cloud Roles

How do attackers grant themselves additional AWS roles and permissions? Adversaries with IAM write permissions attach policies to identities they control, add themselves to privileged groups, modify role trust policies or change IAM Identity Center permission sets — gaining durable, elevated access that looks like normal administration.

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

For leadership

Executive takeaway

Who can change permissions is one of the most important questions in cloud security. That group should be small, monitored and unable to change logging.

How it works in AWS

  1. Attach AdministratorAccess (or an inline policy) to a controlled user or role.
  2. Modify a privileged role’s trust policy to allow an external or controlled principal.
  3. Modify an Identity Center permission set and its assignments.

Sub-technique of T1098 Account Manipulation.

What the attacker needs

  • iam:Attach*Policy / Put*Policy / UpdateAssumeRolePolicy / AddUserToGroup, or sso:* administration

Relevant AWS API calls

AWS API calls relevant to T1098.003
API / eventWhy it matters
iam:AttachRolePolicy / AttachUserPolicyAttach managed policies.
iam:PutRolePolicyAdd inline policies (harder to spot in console views).
iam:UpdateAssumeRolePolicyChange who can assume a role.
iam:CreatePolicyVersionRewrite an existing policy and set it as default.

A risky permission is not an observed attack

Potential exposure

Identities outside the platform team able to modify policies or trust; no permission boundaries on delegated role creation.

Observed behavior

Policy attachments or trust changes outside change windows or not originating from IaC pipeline roles.

Preventive and limiting controls

  • Least-privilege IAM administration
  • Permission boundaries
  • SCPs restricting external trust
  • IaC-only change process for IAM

Mapped cloud controls

  • preventiveLeast-privilege identity policies

    Attaching policies or changing trust requires iam:Attach*/Put*Policy or iam:UpdateAssumeRolePolicy permissions.

  • restrictivePermission boundaries for delegated administration

    If delegated administrators must attach a boundary, new or modified roles cannot exceed the approved maximum.

  • detectiveAlerting on identity and logging changes

    AttachRolePolicy, PutRolePolicy, CreatePolicyVersion and UpdateAssumeRolePolicy are visible in CloudTrail.

See assumptions and validation evidence in the Cloud Control Explorer →

Detection signals and telemetry

  • Attach/Put policy, CreatePolicyVersion, UpdateAssumeRolePolicy — CloudTrail management events
  • New external principals in trust policies — IAM Access Analyzer

MITRE detection strategies (official, cloud-relevant analytics)

  • DET0277 Detection Strategy for Role Addition to Cloud Accounts
    • AN0771 (IaaS): Detection of new IAM roles or policies attached to a user/service in AWS/GCP/Azure outside normal patterns or hours, often following account compromise.Log sources: AWS:CloudTrail — AttachUserPolicy, CreatePolicyVersion, PutRolePolicy
    • AN0772 (Identity Provider): Behavioral chain of a user being granted elevated privileges or roles in Entra ID or Okta following suspicious login or account creation activity.Log sources: m365:audit — Add member to role, Add app role assignment
    • AN0773 (Office Suite): Detection of new admin or role assignment actions within Microsoft 365/O365 environments to elevate access for persistence or lateral movement.Log sources: m365:unified — Add member to role, Set-Mailbox

MITRE mitigations

Official MITRE description (excerpt)

An adversary may add additional roles or permissions to an adversary-controlled cloud account to maintain persistent access to a tenant. For example, adversaries may update IAM policies in cloud-based environments or add a new global administrator in Office 365 environments. With sufficient permissions, a compromised account can gain almost unlimited access to data and settings (including the ability to reset the passwords of other admins).

Common misconceptions

“Inline policies are less important than managed ones.”
Inline policies grant permissions just the same and are often reviewed less.

Attack scenarios using this technique

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