Apply cross-account safeguards with Amazon Bedrock Guardrails enforcements

Plan your guardrail configuration

1. Define your safeguards:

   • Review available guardrail filters in the Amazon Bedrock Guardrails documentation

   • Identify which filter you need. Currently, content filters, denied topics, word filters, sensitive information filters, contextual grounding checks are supported.

   • Important

     Do not include the automated reasoning policy, as it is unsupported for guardrail enforcements and will cause runtime failures.

2. Identify target accounts:

   • Determine which OUs, accounts, or your entire organization will have this guardrail enforced

Create your guardrail in the management account

Create a guardrail in every region where you want to enforce it with one of the following methods:

• Using the AWS Management Console:

  1. Sign in to the AWS Management Console with an IAM identity that has permissions to use the Amazon Bedrock console. Then, open the Amazon Bedrock console at https://eusc-de-east-1.console.amazonaws-eusc.eu/bedrock.

  2. In the left navigation panel, choose Guardrails

  3. Choose Create guardrail

  4. Follow the wizard to configure your desired filters or safeguards (content filters, denied topics, word filters, sensitive information filters, contextual grounding checks)

  5. Do not enable the automated reasoning policy

  6. Complete the wizard to create your guardrail

• Using the API: Use the CreateGuardrail API

Verify

Once created, you should see it in the list of guardrails on the Guardrails landing page or search for it in the list of guardrails using the guardrail name

Create a guardrail version

Create a numeric version to ensure the guardrail configuration remains immutable and cannot be modified by member accounts.

• Using the AWS Management Console:

  1. Select the guardrail created in the previous step in the Guardrails page on the Amazon Bedrock console

  2. Choose Create version

  3. Note the guardrail ARN and the version number (for example, "1", "2")

• Using the API: Use the CreateGuardrailVersion API

Verify

Confirm the version was created successfully by checking the list of versions on the Guardrail detail page.

Attach a resource-based policy

Enable cross-account access by attaching a resource-based policy to your guardrail.

• Using the AWS Management Console – To attach a resource-based policy using the console:

  1. In the Amazon Bedrock Guardrails console, select your guardrail

  2. Choose Add to add a resource-based policy

  3. Add a policy that grants bedrock:ApplyGuardrail permission to all the member accounts or organization. See Share guardrail with your organization in Using resource-based policies for guardrails.

  4. Save the policy

Verify

Test access from a member account using the ApplyGuardrail API to ensure authorization is configured correctly.

Configure IAM permissions in member accounts

Ensure all roles in member accounts have IAM permissions to access the enforced guardrail.

Required permissions

Member account roles need bedrock:ApplyGuardrail permission for the management account's guardrail. See Set up permissions to use Amazon Bedrock Guardrails for detailed IAM policy examples

Verify

Confirm that roles with scoped down permissions in member accounts can successfully call the ApplyGuardrail API with the guardrail.

Enable the Amazon Bedrock Policy Type in AWS Organizations

• Using the AWS Management Console – To enable the Amazon Bedrock policy type using the console:

  1. Navigate to the AWS Organizations console

  2. Choose Policies

  3. Choose Amazon Bedrock policies

  4. Choose Enable Amazon Bedrock policies to enable the Amazon Bedrock policy type for your organization

• Using the API – Use the AWS Organizations EnablePolicyType API with policy type BEDROCK_POLICY

Verify

Confirm the Amazon Bedrock policy type shows as enabled in the AWS Organizations console.

Create and attach an AWS Organizations policy

Create a management policy that specifies your guardrail and attach it to your target accounts or OUs.

• Using the AWS Management Console – To create and attach an AWS Organizations policy using the console:

  1. In the AWS Organizations console, navigate to Policies > Amazon Bedrock policies

  2. Choose Create policy

  3. Specify your guardrail ARN and version

     Important

     Ensure you are specifying the accurate guardrail ARN in the policy. Specifying an incorrect or invalid ARN will result in policy violations, non-enforcement of safeguards, and the inability to use the models in Amazon Bedrock for inference.

  4. Configure selective content guarding controls (optional).

     • Amazon Bedrock APIs allow callers to tag specific content within their input prompts for guardrail evaluation.

     • Selective content guarding controls let administrators decide whether to honor tagging decisions made by API callers.

     • The system and messages controls determine how system prompts and message content are processed by guardrails. Each accepts one of the following values:

       • Selective: Only evaluate content within guard content tags. When no tags are specified, the behavior depends on the control. For system, no content is evaluated, and for messages, all content is evaluated.

       • Comprehensive: Evaluate all content, regardless of guard content tags.

     • If not configured, both controls default to Comprehensive.

     {
         "bedrock": {
             "guardrail_inference": {
                 "us-east-1": {
                     "config_1": {
                         "identifier": {
                             "@@assign": "arn:aws:bedrock:us-east-1:123456789012:guardrail/guardrail-id:1"
                         },
                         "selective_content_guarding": {
                             "system": {
                                 "@@assign": "selective"
                             },
                             "messages": {
                                 "@@assign": "comprehensive"
                             }
                         },
                         "model_enforcement": {
                             "included_models": {
                                 "@@assign": ["ALL"]
                             },
                             "excluded_models": {
                                 "@@assign": ["amazon.titan-embed-text-v2:0", "cohere.embed-english-v3"]
                             }
                         }
                     }
                 }
             }
         }
     }

  5. Save the policy

  6. Attach the policy to your desired targets (organization root, OUs, or individual accounts) by navigating to the Targets tab and choosing Attach

• Using the API – Use the AWS Organizations CreatePolicy API with policy type BEDROCK_POLICY. Use AttachPolicy to attach to targets

Learn more: Amazon Bedrock policies in AWS Organizations

Verify

Check that the policy is attached to the correct targets in the AWS Organizations console.

Test and verify enforcement

Test that the guardrail is being enforced on member accounts.

Plan your guardrail configuration

Define your safeguards

To define your safeguards:

Create your guardrail

Create a guardrail in every region where you want to enforce it.

Via AWS Management Console

To create a guardrail using the console:

1. Sign in to the AWS Management Console with an IAM identity that has permissions to use the Amazon Bedrock console. Then, open the Amazon Bedrock console at https://eusc-de-east-1.console.amazonaws-eusc.eu/bedrock.

2. In the left navigation panel, choose Guardrails

3. Choose Create guardrail

4. Follow the wizard to configure your desired policies (content filters, denied topics, word filters, sensitive information filters)

5. Do not enable the automated reasoning policy

6. Complete the wizard to create your guardrail

Via API

Use the CreateGuardrail API

Verify

Once created, you should see it in the list of guardrails on the Guardrails landing page or search for it in the list of guardrails using the guardrail name

Create a guardrail version

Create a numeric version to ensure the guardrail configuration remains immutable and cannot be modified by member accounts.

Via AWS Management Console

To create a guardrail version using the console:

1. Select the guardrail created in the previous step in the Guardrails page on the Amazon Bedrock console

2. Choose Create version

3. Note the guardrail ARN and the version number (for example, "1", "2")

Via API

Use the CreateGuardrailVersion API

Verify

Confirm the version was created successfully by checking the list of versions on the Guardrail detail page.

Attach a resource-based policy (optional)

If you want to share the guardrail with specific roles in your account, attach a resource-based policy.

Via AWS Management Console

To attach a resource-based policy using the console:

1. In the Amazon Bedrock Guardrails console, select your guardrail

2. Choose Add to add a resource-based policy

3. Add a policy that grants bedrock:ApplyGuardrail permission to the desired roles

4. Save the policy

Enable account-level enforcement

Configure the account to use your guardrail for all Amazon Bedrock invocations. This must be done in every region where you want enforcement.

Via AWS Management Console

To enable account-level enforcement using the console:

1. Navigate to the Amazon Bedrock console

2. Choose Guardrails in the left navigation panel

3. Under the Account-level enforcement configurations section, choose Add

4. Select your guardrail and version

5. Configure selective content guarding controls (optional).

   • Amazon Bedrock APIs allow callers to tag specific content within their input prompts for guardrail evaluation.

   • Selective content guarding controls let administrators decide whether to honor tagging decisions made by API callers.

   • The system and messages controls determine how system prompts and message content are processed by guardrails. Each accepts one of the following values:

     • Selective: Only evaluate content within guard content tags.

     • Comprehensive: Evaluate all content, regardless of guard content tags.

   • If not configured, both controls default to Comprehensive.

6. Submit the configuration

7. Repeat for each region where you want enforcement

Via API

Use the PutEnforcedGuardrailConfiguration API in every region where you want to enforce the guardrail

Verify

You should see the account enforced guardrail under the section Account enforced guardrail configuration on the Guardrails page. You can call ListEnforcedGuardrailsConfiguration API to ensure that the enforced guardrail is listed

Test and verify enforcement

Test using a role in your account

To test enforcement from your account:

1. Make a Amazon Bedrock inference call using InvokeModel, Converse, InvokeModelWithResponseStream, or ConverseStream

2. The account-enforced guardrail should automatically apply to both inputs and outputs

3. Check the response for guardrail assessment information. The guardrail response will include enforced guardrail information.