Cloud Control Explorer
Which AWS controls matter for which attack technique?
A mapping is not a guarantee
A control listed against a technique is relevant under the stated assumptions. It does not mean the technique is fully mitigated, and it does not mean the control is configured or working in your environment. Relationship types: preventive, restrictive, detective, recovery and supporting.T1078.004 Valid Accounts: Cloud Accounts
Securitain guideRelevant controls, why they matter, and the evidence needed to show they work. A relationship means the control is relevant under its assumptions — not that it fully mitigates the technique.
PreventiveMakes the technique infeasible under the stated assumptions.
Phishing-resistant MFA for human privileged access
CTL-IAM-MFAStolen passwords alone cannot complete an interactive sign-in when MFA is enforced for the account.
- Assumptions
- Applies to console / IdP sign-in. Does not protect access keys or already-issued session tokens.
- Evidence required
- IdP or IAM policy showing MFA is required for every privileged human identity
- Credential report showing mfa_active for remaining IAM users
- How to validate
- Review IdP conditional-access settings and the IAM credential report; attempt a test sign-in without MFA in a non-production account.
RestrictiveLimits scope or impact if the technique is attempted.
Replace long-term access keys with temporary credentials
CTL-IAM-NOKEYSShort-lived credentials expire, so a leaked credential is useful for hours rather than indefinitely.
- Assumptions
- Workloads can use roles or federation instead of IAM user keys.
- Evidence required
- Inventory of IAM users with active access keys and their owners
- Justification for each remaining long-term key
- How to validate
- Export the credential report and count active access keys; confirm each is owned and justified.
Find and remove unused credentials and permissions
CTL-IAM-UNUSEDDormant users and keys are attractive because nobody notices when they start being used.
- Assumptions
- Last-used data is reviewed regularly and acted on.
- Evidence required
- Access Analyzer unused-access findings or last-used reports
- Ticket history showing removal of stale credentials
- How to validate
- Compare credential-report last-used dates against the review policy (for example 90 days).
DetectiveProduces evidence that may reveal the technique — if someone acts on it.
Alerting on identity and logging changes
CTL-ALERT-IDENTITYAlerts on unusual sign-ins, root use and new credentials can surface account misuse early.
- Assumptions
- Someone owns and triages the alerts.
- Behavioural detections need baselines; first-use alerts can be noisy.
- Evidence required
- EventBridge / CloudWatch rules for ConsoleLogin anomalies and root activity
- Alert routing and on-call ownership
- How to validate
- Trigger a benign test event (for example a sign-in from a test identity) and confirm it reaches the queue.
Creating CloudWatch alarms for CloudTrail events Review: reviewedManaged threat detection across accounts and regions
CTL-GUARDDUTYGuardDuty can flag credential use from unusual locations or anonymizing infrastructure.
- Assumptions
- GuardDuty is enabled in every account and region used, and findings are monitored.
- Evidence required
- Organization-wide GuardDuty configuration
- Finding export destination and triage records
- How to validate
- Use GuardDuty sample findings to confirm the alert path end to end.
GuardDuty finding types Review: reviewed
Control catalog
Identity & access
Phishing-resistant MFA for human privileged access
CTL-IAM-MFARequire MFA (preferably FIDO2 / passkeys) for every human identity that can reach privileged roles, including the root user, and route workforce access through a federated identity provider rather than IAM users.
AWS: IAM Identity Center, IAM
AWS multi-factor authentication in IAMRelated Securitain capability
Replace long-term access keys with temporary credentials
CTL-IAM-NOKEYSPrefer roles, IAM Identity Center and workload federation (OIDC) over IAM user access keys. Where a key is unavoidable, scope it narrowly, restrict where it can be used from, and rotate it.
AWS: IAM, IAM Identity Center, STS
Manage access keys for IAM usersRelated Securitain capability
Find and remove unused credentials and permissions
CTL-IAM-UNUSEDUse the IAM credential report, last-accessed information and Access Analyzer unused-access findings to retire stale users, keys, roles and permissions on a fixed cadence.
AWS: IAM, IAM Access Analyzer
Generate credential reports for your AWS accountRelated Securitain capability
Least-privilege identity policies
CTL-IAM-LEASTPRIVGrant only the actions and resources a workload or person needs. Treat wildcard actions on IAM, STS, KMS, Secrets Manager, S3 and logging services as high-risk and review them explicitly.
AWS: IAM, IAM Access Analyzer
Constrain iam:PassRole and role-creation permissions
CTL-IAM-PASSROLEScope iam:PassRole to specific role ARNs and use the iam:PassedToService condition so identities cannot hand a more privileged role to a compute service they control.
AWS: IAM, Lambda, EC2, ECS, CloudFormation
Grant a user permissions to pass a role to an AWS serviceRelated Securitain capability
Permission boundaries for delegated administration
CTL-IAM-BOUNDARYWhen teams or pipelines may create roles and users, require a permission boundary so the identities they create can never exceed an approved maximum.
AWS: IAM
Permissions boundaries for IAM entitiesRelated Securitain capability
Scoped cross-account role trust
CTL-TRUST-SCOPETrust specific principals rather than whole accounts, require sts:ExternalId for third parties, and add organization conditions (for example aws:PrincipalOrgID) where trust is meant to stay internal.
AWS: IAM, STS, AWS Organizations
Access to AWS accounts owned by third parties (ExternalId)Related Securitain capability
Governed IAM Identity Center permission sets
CTL-IDC-PERMSETSKeep permission sets least-privilege, limit session duration, separate administrative permission sets, and review account assignments and who can modify permission sets.
AWS: IAM Identity Center
Permission sets in IAM Identity CenterRelated Securitain capability
Federated, scoped CI/CD deployment identities
CTL-CICD-OIDCUse OIDC federation from the CI system instead of stored keys, constrain the trust policy to specific repositories, branches or environments, and give each pipeline stage a separate, least-privilege role.
AWS: IAM, STS
Create an OpenID Connect (OIDC) identity provider in IAMRelated Securitain capability
Organization guardrails
Service control policy guardrails
CTL-ORG-SCPUse SCPs to deny actions no workload should perform — for example stopping organization trails, disabling GuardDuty, leaving the organization, or using unapproved regions — regardless of what IAM policies allow.
AWS: AWS Organizations
Resource control policies and data perimeter
CTL-ORG-RCPUse resource control policies and perimeter conditions so supported resources can only be accessed by identities from your organization, from expected networks.
AWS: AWS Organizations, S3, KMS, Secrets Manager, STS
Resource control policies (RCPs)Related Securitain capability
Logging & detection
Organization-wide, tamper-resistant CloudTrail
CTL-CT-ORGRun a multi-region organization trail that delivers to a dedicated log-archive account, with log file integrity validation, restricted bucket access and retention that matches investigation needs.
AWS: CloudTrail, S3, KMS, AWS Organizations
Selective data-event logging for sensitive resources
CTL-CT-DATAManagement events do not show object reads. Enable data events (for example S3 object-level and Lambda invoke events) for the buckets and functions whose access you would need to reconstruct.
AWS: CloudTrail, S3, Lambda, DynamoDB
Alerting on identity and logging changes
CTL-ALERT-IDENTITYRoute high-signal events — new access keys, policy attachments, trust-policy changes, trail or detector changes, root activity — to an owned alert queue with a documented response.
AWS: EventBridge, CloudWatch, CloudTrail, GuardDuty
T1078.004T1098.001T1098.003T1199T1685.002T1537T1555.006T1136.003T1651T1648
Managed threat detection across accounts and regions
CTL-GUARDDUTYEnable GuardDuty organization-wide, including the protection plans relevant to your workloads (for example S3, EKS and runtime monitoring), and make sure findings reach a monitored destination.
AWS: GuardDuty
Cost and quota anomaly monitoring
CTL-COST-ANOMALYUnexpected compute or AI-service spend is often the first visible sign of hijacked resources. Monitor cost anomalies and keep service quotas close to real need.
AWS: Cost Anomaly Detection, Service Quotas, Budgets
Data protection
S3 Block Public Access at account level
CTL-S3-BPAEnable Block Public Access at the account (or organization) level so a single bucket policy or ACL change cannot make data public.
AWS: S3
Blocking public access to your Amazon S3 storageRelated Securitain capability
Restrictive bucket and resource policies
CTL-S3-POLICYReview resource policies for cross-account and public grants. Restrict sensitive buckets to named principals, organization membership and expected network paths.
AWS: S3, KMS, SQS, SNS, Secrets Manager
Customer managed keys with separated key policies
CTL-KMS-SEPARATIONEncrypt sensitive stores with customer managed KMS keys whose key policies separate key administration from data access, so storage permissions alone are not enough to read data.
AWS: KMS
Managed secrets with scoped access and rotation
CTL-SECRETS-MGMTKeep credentials in Secrets Manager rather than code, environment variables or prompts; scope secretsmanager:GetSecretValue to specific secrets and rotate where supported.
AWS: Secrets Manager, KMS
Secret scanning in code and pipelines
CTL-SECRET-SCANScan repositories, build logs, container images and infrastructure templates for embedded credentials before they ship, and treat any detected AWS key as compromised.
AWS: Repository secret scanning, Third-party scanners
Resilience
Isolated, immutable backups with tested restores
CTL-BACKUP-ISOLATIONKeep backups in a separate account with vault lock or Object Lock so a compromised workload identity cannot delete them, and test restoration on a schedule.
AWS: AWS Backup, S3 Object Lock
Credential compromise response runbook
CTL-IR-RUNBOOKDocument who deactivates keys, revokes active role sessions, preserves evidence and communicates — and rehearse it. Revoking a key does not end sessions already issued from it.
AWS: IAM, STS, CloudTrail
Workload identity
Enforce IMDSv2 and limit metadata reachability
CTL-IMDSV2Require IMDSv2 session tokens and use a hop limit of 1 where containers should not reach the node metadata service, reducing credential theft through SSRF or container escape.
AWS: EC2, EKS
Per-workload EKS identities
CTL-EKS-IDENTITYUse EKS Pod Identity or IRSA so each service account receives only its own role, keep the node role minimal, and stop pods from inheriting node credentials.
AWS: EKS, IAM
Learn how EKS Pod Identity grants pods access to AWS services
Least-privilege cluster access and audit logging
CTL-EKS-RBACGrant cluster access through EKS access entries and narrowly scoped Kubernetes RBAC, avoid broad cluster-admin bindings, and enable control plane audit logs.
AWS: EKS
Grant IAM users access to Kubernetes with EKS access entries
Restricted pod security settings
CTL-POD-HARDENINGApply the Kubernetes "restricted" Pod Security Standard (no privileged containers, host namespaces or hostPath mounts) so a compromised container cannot trivially reach the node.
AWS: EKS, ECS
AI agent security
Dedicated least-privilege identity per agent and tool
CTL-AI-ROLEGive each agent tool its own narrowly scoped role (or a session policy) instead of a shared broad role, so a manipulated agent can only do what that tool legitimately needs.
AWS: IAM, STS, Amazon Bedrock, Lambda
Authorization enforced by tools, not the model
CTL-AI-TOOL-AUTHZTools must check the end user’s own permissions (and tenant) on every call. A model’s decision to call a tool is never an authorization decision.
AWS: API Gateway, Lambda, Amazon Bedrock
Human approval for consequential agent actions
CTL-AI-HITLRequire an out-of-band human confirmation that shows the exact parameters before an agent deletes, transfers, grants access, spends money or contacts external parties.
AWS: Step Functions, Amazon Bedrock
MITRE ATLAS mitigation: Human In-the-Loop for AI Agent Actions
Treat retrieved and tool-returned content as untrusted
CTL-AI-UNTRUSTEDLabel provenance, keep untrusted content separate from instructions, restrict which tools can run after untrusted input is read, and control who can write to knowledge sources.
AWS: Amazon Bedrock, S3
MITRE ATLAS mitigation: Restrict AI Agent Tool Invocation on Untrusted Data
Input and output guardrails (defense in depth)
CTL-AI-GUARDRAILSFilters for prompt attacks, sensitive data and unsafe output reduce risk but are probabilistic. Use them alongside — never instead of — authorization and least privilege.
AWS: Amazon Bedrock Guardrails
Detect and filter harmful content by using Amazon Bedrock Guardrails
Auditable agent and model activity
CTL-AI-LOGGINGRecord which user and agent triggered each tool call, the parameters, the result and the content sources used, alongside model invocation logs and CloudTrail.
AWS: Amazon Bedrock, CloudTrail, CloudWatch
Monitor model invocation using CloudWatch Logs and Amazon S3
Restricted agent egress and connectors
CTL-AI-EGRESSLimit the domains and services an agent runtime can reach, avoid generic "fetch any URL" tools, and approve third-party connectors before use.
AWS: VPC, VPC endpoints, API Gateway
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