MITRE ATT&CK describes Valid Accounts: Cloud Accounts (T1078.004) as the misuse of legitimate cloud identities for access, persistence, higher privileges or stealth. The account may belong to a person, service or administrator in an IaaS platform, identity provider, office suite or SaaS application. A successful sign-in is evidence of account use, not by itself proof of who controlled it.
How it works
MITRE places this sub-technique under Valid Accounts (T1078). Cloud identities may be native to a service or linked to an on-premises identity through sync or federation. An attacker with compromised credentials for a user or service account can make authorized-looking requests within that account’s permissions. Brute Force (T1110) is one possible way to obtain access; it is a separate technique.
What Signals Point to T1078.004 Cloud Accounts Misuse?
MITRE’s cloud account detection strategy compares sign-ins with usual locations, protocols and account behavior. It also looks for API calls beyond a principal’s normal scope, such as new IAM changes or access to services that principal has not used before. In office suites, a sudden rise in downloads or external sharing can add context. Check the account’s role and usual workload before treating any single event as misuse.
Which Records Should an Investigator Join?
Start with identity provider sign-ins, the principal, source network, device, authentication method and session history. Join those records to cloud audit events for role changes, API access and file activity. Google recommends checking new passkeys and MFA devices, OAuth grants, browser sessions and refresh tokens when account compromise is suspected. Elastic’s AWS investigation shows why CloudTrail events from one external source can matter even when individual alerts look routine.
A real example
Elastic reported stolen AWS credentials used from an external IP across user and instance role sessions and linked otherwise separate CloudTrail alerts by source IP. In another case, Google Threat Intelligence Group and Mandiant attributed cloud identity misuse to Star Blizzard (COLD RELIC). Google mapped MFA backup code authentication to T1078.004 and described suspicious Google Workspace and Entra ID account activity. These are distinct cases, not a sequence every intrusion follows.
Who uses it
How to detect
-
Review abnormal sign-ins
MITRE DET0546 calls for comparing privileged sign-ins with usual location, protocol and user patterns. Google adds MFA backup code logins from unexpected networks to the review queue.
-
Correlate cloud API calls
MITRE AN1504 flags IAM changes or services outside an account's normal scope. Elastic recommends joining CloudTrail events by principal, source IP and time to connect otherwise separate AWS alerts.
-
Check cloud productivity activity
MITRE AN1505 looks for unusual downloads and external sharing. Compare office suite audit events with the user's normal work and the timing of the suspicious session.
-
Inspect identity persistence
Google recommends reviewing newly enrolled passkeys or MFA devices, third-party OAuth grants and refresh tokens after a suspicious cloud sign-in.
How to defend
-
Require stronger authentication
MITRE M1032 recommends MFA for cloud accounts, especially privileged ones. Google recommends phishing-resistant MFA and tighter approval of new authenticators.
-
Limit privileged access
MITRE M1026 recommends routine permission reviews and checks for unauthorized privileged accounts. Use temporary privilege where the cloud platform supports it.
-
Contain affected AWS identities
For the AWS case it investigated, Elastic recommends rotating affected credentials, revoking active sessions and checking whether instance role credentials can be reissued from a compromised host.
-
Revoke persistent cloud access
For suspected Google Workspace or Entra ID account compromise, Google recommends revoking browser sessions, refresh tokens and unauthorized OAuth grants, then removing rogue passkeys, backup codes and MFA devices.
Frequently Asked Questions
Is T1078.004 the same as stealing a password?
No. MITRE uses T1078.004 for misuse of an existing cloud account. Brute Force (T1110) or phishing may precede that use, but each describes a different action.
Does a successful cloud sign-in prove malicious use?
No. Compare the sign-in with the account's normal device, network, authentication method and later API or file activity before assigning an outcome.
Why check MFA backup codes and passkeys?
Google mapped MFA backup code authentication to T1078.004 in its COLD RELIC case study. It also observed attacker-enrolled passkeys and recommends checking for unauthorized authenticators during response.
Sources
- Valid Accounts: Cloud Accounts, T1078.004MITRE ATT&CK
- Valid Accounts, T1078MITRE ATT&CK
- 56 AWS API calls in 141 milliseconds: Inside an AI-run intrusionElastic Security Labs
- COLD RELIC: A Case Study in Impersonation and CompromiseGoogle Threat Intelligence Group and Mandiant
