Evaluating Configurations for Reducing Problematic Emotional Engagement in Enterprise LLM Deployments: Implications for Insider Threat Risk
The risks of Large Language Models (LLMs) include triggering psychological drivers associated with malicious insider threat behavior. This study utilized AWS Bedrock to demonstrate that specific system-level configurations and guardrails can effectively mitigate these risks by reducing problematic...
