MITRE ATT&CK technique · T1199
Trusted Relationship
How do attackers abuse trusted relationships with vendors and other AWS accounts? Adversaries compromise a party you already trust — a vendor, MSP or another account — and use the access you granted that party.
For leadership
Executive takeaway
How it works in AWS
- Compromise a trusted party’s environment or platform.
- Assume the cross-account role your environment exposes to them.
- Operate with that role’s permissions.
What the attacker needs
- A role trusting the third party
- Compromise of that party
Relevant AWS API calls
| API / event | Why it matters |
|---|---|
| sts:AssumeRole | Cross-account entry (logged in the role’s account). |
A risky permission is not an observed attack
Trust of account root without ExternalId or conditions; broad permissions; no owner.
AssumeRole from external accounts at unusual times or from new sources.
Preventive and limiting controls
- Specific principals + ExternalId
- Least privilege for vendor roles
- Decommission with the contract
Mapped cloud controls
- preventiveScoped cross-account role trust
Specific principals plus ExternalId (third parties) or organization conditions (internal) prevent unexpected principals from assuming the role.
- restrictiveLeast-privilege identity policies
A trusted third party should receive only the permissions its service needs, limiting impact if it is compromised.
- detectiveAlerting on identity and logging changes
AssumeRole events from external accounts can be baselined; new source accounts or unusual times stand out.
See assumptions and validation evidence in the Cloud Control Explorer →
Detection signals and telemetry
- External AssumeRole outside allow-list — CloudTrail
MITRE detection strategies (official, cloud-relevant analytics)
- DET0488 Detect abuse of Trusted Relationships (third-party and delegated admin access)
- AN1347 (Identity Provider): Behavioral chain: (1) delegated admin or external identity establishes session (e.g., partner/reseller DAP, B2B guest, SAML/OAuth trust); (2) role elevation or app consent/permission grant; (3) downstream privileged actions in the tenant. Correlate IdP sign-in, admin/role assignment, and consent/admin-on-behalf events.Log sources: azure:signinlogs — InteractiveUser, ServicePrincipalSignIn; azure:audit — Add delegated admin / Assign admin roles / Update application consent; m365:unified — Set-PartnerOfRecord / CompanyAdministrator role assignments / New-DelegatedAdminRelationship
- AN1348 (IaaS): Behavioral chain: (1) cross-account or third-party principal assumes a role into the tenant/subscription/project; (2) privileged API calls are made in short succession; (3) access originates from unfamiliar networks or geos. Correlate assume-role/federation events with sensitive API usage.Log sources: AWS:CloudTrail — AssumeRole,AssumeRoleWithSAML,AssumeRoleWithWebIdentity; AWS:CloudTrail — CreateUser|AttachRolePolicy|CreateAccessKey|UpdateAssumeRolePolicy|CreateLoginProfile; gcp:audit — google.iam.credentials.generateAccessToken / serviceAccountTokenCreator
- AN1349 (SaaS): Behavioral chain: (1) third-party app or admin connects via OAuth/marketplace install; (2) high-privilege scopes granted; (3) anomalous actions (mass read/exports, admin changes).Log sources: saas:googleworkspace — OAuth2 authorization grants / Admin role assignments; saas:salesforce — ConnectedApp OAuth policy change / Login as user
- AN1350 (Office Suite): Behavioral chain: (1) delegated administration offers/relationships created or modified by partner tenants; (2) mailbox delegation/impersonation enabled; (3) follow-on access from partner IPs.Log sources: m365:unified — Add-DelegatedAdmin, Set-PartnerOfRecord, Add-MailboxPermission, Set-OrganizationRelationship; azure:signinlogs — InteractiveUser, NonInteractiveUser
MITRE mitigations
Official MITRE description (excerpt)
Adversaries may breach or otherwise leverage organizations who have access to intended victims. Access through trusted third party relationship abuses an existing connection that may not be protected or receives less scrutiny than standard mechanisms of gaining access to a network.
Common misconceptions
- “ExternalId is a password.”
- It prevents confused-deputy misuse; it does not protect you if the vendor itself is compromised.
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