Amazon
youtube
Amazon Bedrock Guardrails is a comprehensive solution that helps organizations implement customizable safety, privacy, and truthfulness safeguards across their generative AI applications. As the only responsible AI capability of its kind from a major cloud provider, it works consistently across multiple foundation models to prevent hallucinations, block harmful content, and protect sensitive information.
Using Automated Reasoning—a first in generative AI—it prevents factual errors from hallucinations, blocks up to 88% of undesirable and harmful multimodal content, and filters over 75% of hallucinated responses using relevance and contextual grounding checks. Amazon Bedrock Guardrails works with a wide range of models, including FMs supported in Amazon Bedrock, fine-tuned models, self-hosted models, and 3P models outside of Amazon Bedrock.
|
2025/04/08
youtube
![[PyCon APAC 2025報告会] フィリピンのマニラに行ってきました！参加メンバーによる報告会](https://i.ytimg.com/vi/Zyjq5Dwj3dk/sddefault.jpg)
