Skip to main content

Amazon Bedrock

Amazon Bedrock security, privacy, and responsible AI

Deploy generative AI with confidence using enterprise-grade security, privacy, and responsible AI controls

Develop generative AI on a trusted foundation

Amazon Bedrock provides enterprise-grade security, privacy, and responsible AI safeguards that enable organizations to build generative AI applications with confidence. Your data remains completely secure and private—inputs and outputs are never shared with model providers or used to train base models. Built-in protections include configurable content safeguards, personally identifiable information (PII) detection, and industry-first Automated Reasoning checks that helps prevent hallucinations, providing comprehensive safeguards against harmful content and AI risks. 

Missing alt text value

Design AI responsibly with built-in safety and guardrails

From data protection to hallucination mitigation—comprehensive enterprise security, privacy, and responsible AI controls for trusted AI deployment

Maintain full control over your data

Amazon Bedrock never shares your data with model providers or uses it to train foundation models. When you tune a foundation model, we create a private copy specifically for your use, keeping your data separate and secure. With full encryption in transit and at rest, AWS Key Management Service (KMS), and private VPC connectivity through AWS PrivateLink, your most sensitive information remains protected. 

Missing alt text value

Build trusted AI applications with configurable safeguards

Amazon Bedrock Guardrails helps you build generative AI applications at scale with customizable safeguards that help filter harmful content, detect hallucinations, and redact sensitive information. With industry-leading multi-modal toxicity detection that blocks up to 88% of harmful content, you can filter hate speech, insults, sexual content, violence, misconduct, and prompt attacks using configurable thresholds that match your specific needs. Guardrails also helps you detect and redact personally identifiable information (PII) in user inputs and model responses, with options to select from predefined PII types or define custom sensitive-information types using regular expressions. Deploy these protections across any foundation model of your choice —whether using Amazon Bedrock models, third-party options (OpenAI, Google Gemini), or agent frameworks like Strands Agents—maintaining consistent safety, privacy, and truthfulness controls across all your generative AI applications. 

Improve AI reliability, accuracy, and explainability with automated reasoning

Automated Reasoning checks in Amazon Bedrock Guardrails is the first and only generative AI safeguard that helps prevent factual errors from hallucinations. Using sound mathematical techniques to verify, correct, and logically explain generated information, it identifies correct model responses with up to 99% accuracy and provides provable explanations for why the information is accurate. Contextual grounding checks filter over 75% of hallucinated responses for Retrieval Augmented Generation (RAG) and summarization workloads by validating outputs against source information. Both capabilities help you build more trustworthy AI applications by improving output reliability, accuracy, and explainability. 

Deploy AI with enterprise governance

Amazon Bedrock adheres to key industry regulations including GDPR, HIPAA, SOC, and FedRAMP High, helping organizations deploy AI in regulated environments with confidence. With fine-grained access controls through IAM policies, comprehensive audit trails via CloudTrail, and advanced observability through CloudWatch AppSignals, you can maintain governance and meet compliance requirements. Amazon Bedrock provides the security certifications and controls needed for regulated industries while enabling seamless integration with existing GRC (Governance, Risk, and Compliance) systems. 

Missing alt text value

Did you find what you were looking for today?

Let us know so we can improve the quality of the content on our pages