aws·Apr 3, 2026·2 min readCurated summary

Amazon Bedrock Guardrails supports cross-account safeguards with centralized control and management | Amazon Web Services

Amazon Bedrock Guardrails now supports cross-account safeguards, allowing organizations to centrally enforce safety controls across AWS accounts and organizational units. Administrators can apply immutable, versioned guardrails to all Bedrock model invocations while still allowing account- or application-specific policies. The capability is generally available across commercial and GovCloud Regions where Bedrock Guardrails is supported. ## Centralized Organization- and Account-Level Enforcement - **Organization-level enforcement** uses an Amazon Bedrock policy created in the AWS Organizations management account. - Policies can attach a specified guardrail and version to: - The organization root - Organizational units - Individual AWS accounts - The selected guardrail is automatically applied to Bedrock inference requests across targeted member entities. - Different policies and guardrails can be assigned to different accounts or organizational units. - **Account-level enforcement** applies a configured guardrail to all Bedrock inference API calls within one account and Region. ## Configuring Guardrail Coverage - Guardrails must use a specific version so their configuration remains immutable and cannot be changed by member accounts. - Administrators can choose whether enforcement: - Includes or excludes specific Bedrock models - Covers all or only selected system and user prompt content - **Comprehensive** mode guards all content, regardless of caller-provided tags. - **Selective** mode relies on callers to identify content requiring protection, reducing processing for pre-validated inputs. ## Testing and Verification - Account-level enforcement can be configured in the Amazon Bedrock Guardrails console. - Enforcement can be tested with: - `InvokeModel` - `InvokeModelWithResponseStream` - `Converse` - `ConverseStream` - Responses include guardrail assessment details and identify the enforced guardrail. - Member accounts can verify organization-level enforcement in the Bedrock console. ## Important Considerations - Organizations must meet prerequisites such as configuring resource-based policies for guardrails. - Incorrect or invalid guardrail ARNs can cause policy violations, prevent safeguards from being enforced, and block model inference. - Automated Reasoning checks are not supported. - Charges apply for each enforced guardrail based on its configured safeguards. ## Availability Cross-account safeguards are generally available in all commercial and GovCloud AWS Regions where Amazon Bedrock Guardrails is available. Organizations can enable the feature through the Amazon Bedrock and AWS Organizations consoles. Overall, the capability gives security teams a centralized way to enforce responsible AI requirements while reducing the need to audit guardrail settings independently in every account and application.

Read original(opens in new tab)

Ai SafetyAmazon BedrockAws OrganizationsGuardrail+4

google·Aug 11, 2025Original article

Enabling physician-centered oversight for AMIE (opens in new tab)

Guardrailed-AMIE (g-AMIE) is a diagnostic AI framework designed to perform patient history-taking while strictly adhering to safety guardrails that prevent it from providing direct medical advice. By decoupling data collection from clinical decision-making, the system enables an asynchronous oversight model where primary care physicians (PCPs) review and finalize AI-generated medical summaries. In virtual clinical trials, g-AMIE’s diagnostic outputs and patient communications were preferred by overseeing physicians and patient actors over human-led control groups. ## Multi-Agent Architecture and Guardrails * The system utilizes a multi-agent setup powered by Gemini 2.0 Flash, consisting of a dialogue agent, a guardrail agent, and a SOAP note agent. * The dialogue agent conducts history-taking in three distinct phases: general information gathering, targeted validation of a differential diagnosis, and a conclusion phase for patient questions. * A dedicated guardrail agent monitors and rephrases responses in real-time to ensure the AI abstains from sharing individualized diagnoses or treatment plans directly with the patient. * The SOAP note agent employs sequential multi-step generation to separate summarization tasks (Subjective and Objective) from more complex inferential tasks (Assessment and Plan). ## The Clinician Cockpit and Asynchronous Oversight * To facilitate human review, researchers developed the "clinician cockpit," a web interface co-designed with outpatient physicians through semi-structured interviews. * The interface is structured around the standard SOAP note format, presenting the patient’s perspective, measurable data, differential diagnosis, and proposed management strategy. * This framework allows overseeing PCPs to review cases asynchronously, editing the AI’s proposed differential diagnoses and management plans before sharing a final message with the patient. * The separation of history-taking from decision-making ensures that licensed medical professionals retain ultimate accountability for patient care. ## Performance Evaluation via Virtual OSCE * The system was evaluated in a randomized, blinded virtual Objective Structured Clinical Examination (OSCE) involving 60 case scenarios. * g-AMIE’s performance was compared against primary care physicians, nurse practitioners, and physician assistants who were required to operate under the same restrictive guardrails. * Overseeing PCPs and independent physician raters preferred g-AMIE’s diagnostic accuracy and management plans over those of the human control groups. * Patient actors reported a preference for the messages generated by g-AMIE compared to those drafted by human clinicians in the study. While g-AMIE demonstrates high potential for human-AI collaboration in diagnostics, the researchers emphasize that results should be interpreted with caution. The workflow was specifically optimized for AI characteristics, and human clinicians may require specialized training to perform effectively within such highly regulated guardrail frameworks.

Ai SafetyAIGenerative AINatural Language Processing+4