返回产品列表
B

产品详情

Bedrock Guardrail Firewall

Privacy-first guardrails for generative AI

发布日期 · UTC
2026-11-01
产品域名
—
域名后缀
—
产品方向
人工智能

关于产品

Bedrock Guardrail Firewall is an open-source, privacy-first policy enforcement layer for generative AI. It redacts sensitive data locally, detects prompt attacks, checks grounding and authorization, and can optionally call Amazon Bedrock Guardrails or Microsoft Presidio. Offline by default, fail-closed in production, and backed by metadata-only, tamper-evident audit trails.

关键词

ai safetydata maskingprompt injection