Amazon Bedrock Guardrails expands support for code domain

This article announces expanded code domain support for Amazon Bedrock Guardrails, a safety service for generative AI applications.

• Bedrock Guardrails now protects code elements including prompts, comments, variables, and string literals
• Detects harmful content across five categories: misconduct, violence, sexual, insult, and hate
• Prevents prompt injection attacks, jailbreaks, and system prompt leakage attempts
• Supports twelve major programming languages including Python, JavaScript, Java, and C++
• Customizable denied topics filters allow organization-specific code restrictions
• Enhanced PII detection optimized for coding contexts, code snippets, and hybrid content
• Available in Standard Tier with cross-Region inference support

Amazon Bedrock Guardrails now provides comprehensive safety controls for AI-assisted code development, protecting against malicious code generation and data exfiltration while maintaining developer productivity.