Amazon Bedrock Guardrails announces tiers for content filters and denied topics

Amazon Bedrock Guardrails has introduced a new Standard tier for content filtering and content control, enhancing AI safety features across multiple dimensions.

• Offers improved content filtering with better contextual understanding
• Supports detection of content variations, including typographical errors
• Expands language support to 60 languages
• Provides safeguards against harmful content, prompt attacks, and model hallucinations
• Can be applied across foundation models, including self-hosted and third-party models
• Enables detection and blocking of undesirable content
• Helps redact personally identifiable information (PII)

The Standard tier requires explicit opt-in for cross-region inference and provides more robust defense against various content filtering challenges.