Amazon Bedrock Guardrails gets automated policy refinement
Amazon Bedrock Guardrails now offers automated refinement workflows for its Automated Reasoning checks, simplifying policy improvement for generative AI responses. These new features help customers reduce hallucinations and improve validation accuracy with less manual effort by iteratively refining policies based on test cases or resolving ambiguities in variable descriptions. Both workflows are available via API and the AWS Management Console in all regions supporting Guardrails Automated Reasoning checks.
- →Iterative policy improvement workflow
- →Ambiguity reduction workflow
- →New Automated Reasoning checks refinement workflows for Bedrock Guardrails
- →Availability and access for new workflows
Features (2) ›
- Iterative policy improvement workflow
The iterative policy improvement workflow allows customers to provide natural language tests for a policy, enabling the system to automatically deduce and suggest necessary policy changes to pass those tests.
- Ambiguity reduction workflow
The resolve policy ambiguities workflow helps customers who experience frequent ambiguous translation results by automatically refining variable descriptions and type definitions, thereby reducing the occurrence of such ambiguities.
Notes (2) ›
- New Automated Reasoning checks refinement workflows for Bedrock Guardrails
AWS announces new automated refinement workflows for Automated Reasoning checks in Amazon Bedrock Guardrails, aiming to improve policy definition and validation accuracy for generative AI. These workflows help customers refine policies with less manual effort, leading to more reliable results.
- Availability and access for new workflows
Both new refinement workflows are accessible through the Amazon Bedrock APIs and the AWS Management Console. Customers can initiate a workflow by selecting 'Refine policy' on the policy page.
https://aws.amazon.com/about-aws/whats-new/2026/06/amazon-bedrock-guardrails/