Type
Merged detection coverage for AWS Bedrock Claude abuse patterns, including prompt injection, sensitive-data exposure, high-risk tool invocation, cross-region inference, hostile prompts, unusually large prompts, and excessive token output anomalies.