Every account in an AWS Organization should be covered by authorization guardrails
This check looks at whether service control policies are turned on in your AWS Organization and attached so that member accounts inherit them. The safe setting is to have guardrail policies in place that set the outer limit of what every member account is allowed to do.
Why it matters
Service control policies set the maximum permissions for accounts in an organization. Even an administrator inside a member account cannot go beyond what the policies allow, so they are a reliable place to block actions you never want, such as leaving the organization or turning off logging. Without them, each account's own IAM settings are the only barrier.
Typical guardrails include preventing accounts from leaving the organization, stopping changes to audit logging, and limiting which Regions can be used. They complement, rather than replace, the permissions you grant inside each account.
AWS explains how they work in Service control policies (SCPs). Note that they do not apply to the management account itself.
How to check it yourself
Console path: AWS Organizations, Policies, Service control policies. Confirm the policy type is enabled, then open a policy and review its Targets tab to see which roots, organizational units and accounts it covers.
Read-only AWS CLI commands from the management account:
aws organizations list-roots --query "Roots[].{id:Id,policyTypes:PolicyTypes}"
aws organizations list-policies --filter SERVICE_CONTROL_POLICY --query "Policies[].{id:Id,name:Name,aws:AwsManaged}" -o table
aws organizations list-targets-for-policy --policy-id <policy-id> --query "Targets[].{name:Name,type:Type}" -o tableThese only list configuration. A custom policy attached at the root, or at every organizational unit, means all member accounts are covered. A policy that only allows everything does not add a guardrail.
How to fix it
Console steps: in AWS Organizations, Policies, Service control policies, choose Enable if needed, create a policy that denies the actions you want to prevent, then attach it to the root or to organizational units.
Script (replace the placeholders; try the policy on a test organizational unit first):
aws organizations enable-policy-type --root-id <root-id> --policy-type SERVICE_CONTROL_POLICY
aws organizations create-policy --name deny-leave-organization --type SERVICE_CONTROL_POLICY --description "Members cannot leave the organization" --content '{"Version":"2012-10-17","Statement":[{"Effect":"Deny","Action":"organizations:LeaveOrganization","Resource":"*"}]}'
aws organizations attach-policy --policy-id <policy-id> --target-id <ou-id>ConfigCobra itself only reads your AWS accounts and shows the fix; it never applies it.
What changes when you fix it
A deny policy takes effect for every principal in the targeted accounts straight away, including roles that automation uses. If a pipeline, a vendor integration or an administrator depends on an action you block, that call starts failing with an access denied error. Review CloudTrail history for the actions you plan to deny, roll the policy out to a test organizational unit first, and keep a documented way to detach it in an emergency. Teams will also need to know that a permission they can see in IAM may still be blocked from above.
ConfigCobra's remediation impact analysis shows the affected identities and resources before you apply the fix.
Included in
| Benchmark | CIS ID | Profile |
|---|---|---|
| CIS Amazon Web Services Foundations v7.0.0 | 2.1.2 | Level 2 |
Check this on your AWS
ConfigCobra runs this check with every other control of the CIS Amazon Web Services Benchmarks, read-only.