MITRE ATT&CK technique · T1530
Data from Cloud Storage
How do attackers steal data from Amazon S3? Adversaries read data from cloud object storage using either misconfigured public / cross-account access or permissions held by a compromised identity. The reads are ordinary GetObject calls.
For leadership
Executive takeaway
How it works in AWS
- Find buckets through discovery or public references.
- List and read objects with available permissions.
- Prefer backups, exports and data-lake prefixes for maximum value.
What the attacker needs
- s3:GetObject via identity or resource policy
- kms:Decrypt if objects use customer managed keys
Relevant AWS API calls
| API / event | Why it matters |
|---|---|
| s3:ListBucket | List object keys. |
| s3:GetObject | Read object content (data event). |
A risky permission is not an observed attack
Public or cross-account bucket grants; identities with s3:* on all buckets.
Bulk GetObject by an identity, access from unknown accounts or anonymous principals.
Preventive and limiting controls
- Account-level Block Public Access
- Restrictive bucket policies and data perimeter
- SSE-KMS with scoped key policies
Mapped cloud controls
- preventiveS3 Block Public Access at account level
Block Public Access stops reads that rely on public bucket policies or ACLs, including anonymous internet access and cross-account access granted through a public policy.
- restrictiveRestrictive bucket and resource policies
Bucket policies that restrict principals and require organization membership limit who can read data even with valid credentials elsewhere.
- restrictiveCustomer managed keys with separated key policies
With SSE-KMS customer managed keys, reading objects also requires kms:Decrypt, adding a second, separately governed permission.
- detectiveSelective data-event logging for sensitive resources
Object-level reads are invisible without S3 data events; with them, bulk GetObject activity can be reconstructed and alerted on.
- detectiveManaged threat detection across accounts and regions
GuardDuty S3 Protection analyses data events for anomalous access patterns.
See assumptions and validation evidence in the Cloud Control Explorer →
Detection signals and telemetry
- Object-level access anomalies — S3 data events; GuardDuty S3 Protection
- External access findings — IAM Access Analyzer
MITRE detection strategies (official, cloud-relevant analytics)
- DET0484 Multi-Platform Cloud Storage Exfiltration Behavior Chain
- AN1328 (IaaS): Spike in object access from new IAM user or role followed by data exfiltration to external IPsLog sources: AWS:CloudTrail — GetObject, CopyObject; AWS:CloudTrail — AssumeRole; AWS:VPCFlowLogs — Unusual volume of data transferred from S3 storage endpoints to non-corporate IPs
- AN1329 (SaaS): OAuth token granted to external app followed by download of high-volume files in OneDrive/Google DriveLog sources: m365:unified — FileAccessed, FileDownloaded, ConsentGranted
- AN1330 (Office Suite): Internal user account accesses shared links outside org followed by mass file downloadLog sources: m365:sharepoint — AnonymousLinkCreated, FileDownloaded; azure:signinlogs — SigninSuccess
MITRE mitigations
Official MITRE description (excerpt)
Adversaries may access data from cloud storage.
Many IaaS providers offer solutions for online data object storage such as Amazon S3, Azure Storage, and Google Cloud Storage. Similarly, SaaS enterprise platforms such as Office 365 and Google Workspace provide cloud-based document storage to users through services such as OneDrive and Google Drive, while SaaS application providers such as Slack, Confluence, Salesforce, and Dropbox may provide cloud storage solutions as a peripheral or primary use case of their platform.
Common misconceptions
- “Encryption at rest stops data theft.”
- SSE-S3 decrypts transparently for anyone with read access; only KMS key policies add a separate control.
Attack scenarios using this technique
Compromised AWS access key
Dangerous cross-account role trust
Excessive IAM permissions
Privilege escalation through IAM policy relationships
Compromised deployment pipeline role
Exposed machine identity credentials
Sensitive S3 data exposure
CloudTrail logging impairment
Sensitive data access through an overprivileged workload role
Compromised EKS service account
Container-to-cloud identity exposure
AI agent with excessive AWS permissions
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