MITRE ATT&CK technique · T1528
Steal Application Access Token
How are application and workload access tokens stolen? Adversaries steal tokens that applications and workloads use to call APIs — Kubernetes service account tokens, OAuth tokens, CI/CD tokens — and replay them.
For leadership
Executive takeaway
How it works in AWS
- Read mounted tokens from a compromised pod or process.
- Trick users into granting OAuth consent to a malicious app.
- Harvest tokens from CI variables or agent configuration.
What the attacker needs
- Access to where tokens are stored or issued
Relevant AWS API calls
| API / event | Why it matters |
|---|---|
| sts:AssumeRoleWithWebIdentity | Workload tokens exchanged for AWS credentials (IRSA). |
A risky permission is not an observed attack
Shared service accounts, long-lived tokens, broad roles mapped to workloads.
Token use from outside the workload; unexpected AssumeRoleWithWebIdentity sources.
Preventive and limiting controls
- Per-workload identities
- Disable token automount where unused
- Short token lifetimes
Mapped cloud controls
- restrictivePer-workload EKS identities
Per-service-account roles mean a stolen pod token only yields that workload’s AWS permissions.
- preventiveSecret scanning in code and pipelines
Tokens and keys committed to code or logs are a common source of stolen credentials.
See assumptions and validation evidence in the Cloud Control Explorer →
Detection signals and telemetry
- Workload tokens used outside the cluster — CloudTrail; GuardDuty EKS protection
MITRE detection strategies (official, cloud-relevant analytics)
- DET0515 Detection Strategy for T1528 - Steal Application Access Token
- AN1423 (Containers): Access and retrieval of container service account tokens followed by unauthorized API requests using those tokens to interact with the Kubernetes API server or internal services.Log sources: kubernetes:audit — GET or LIST requests to /var/run/secrets/kubernetes.io/serviceaccount/ followed by access to the Kubernetes API server
- AN1424 (IaaS): Token retrieval from instance metadata endpoints such as AWS IMDS or Azure IMDS, followed by API usage using the obtained token from non-standard applications.Log sources: AWS:CloudTrail — GetInstanceIdentityDocument or IMDSv2 token requests; AWS:CloudTrail — Use of temporary credentials issued from IMDS access
- AN1425 (Identity Provider): Unusual OAuth app registration followed by user-granted OAuth tokens and subsequent high-privilege resource access via those tokens.Log sources: azure:audit — App registrations or consent grants by abnormal users or at unusual times
- AN1426 (Office Suite): Use of OAuth tokens by third-party apps to access user mail, calendar, or SharePoint resources where the token was granted recently or via spearphishing.Log sources: m365:unified — App-only or delegated access patterns where client_id != known enterprise apps
- AN1427 (SaaS): Programmatic access to user content via stolen access tokens in platforms like Slack, GitHub, Google Workspace — especially from new IPs, apps, or excessive resource access.Log sources: saas:googleworkspace — Access via OAuth credentials with unusual scopes or from anomalous IPs; saas:slack — OAuth token use by unknown app client_id accessing private channels or files
MITRE mitigations
Official MITRE description (excerpt)
Adversaries can steal application access tokens as a means of acquiring credentials to access remote systems and resources.
Application access tokens are used to make authorized API requests on behalf of a user or service and are commonly used as a way to access resources in cloud and container-based applications and software-as-a-service (SaaS). Adversaries who steal account API tokens in cloud and containerized environments may be able to access data and perform actions with the permissions of these accounts, which can lead to privilege escalation and further compromise of the environment.
Common misconceptions
- “Tokens expire quickly so theft is harmless.”
- Short lifetimes limit duration, not what can be done in that window.
AI security
Related MITRE ATLAS techniques
ATLAS and ATT&CK are separate frameworks. These relationships are listed explicitly, with their basis.
- AML.T0098 AI Agent Tool Credential HarvestingRelated AI-specific behavior (Securitain analysis).
Attack scenarios using this technique
Sensitive data access through an overprivileged workload role
Compromised EKS service account
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