ShED-HD introduces a novel lightweight framework for detecting hallucinations in Large Language Models without requiring additional computational resources.

• Leverages Shannon entropy distributions to identify factual inconsistencies in LLM outputs
• Achieves comparable accuracy to more resource-intensive methods while requiring significantly less computation
• Designed specifically for edge devices with limited processing power
• Enables real-time hallucination detection for security-critical applications

Security Impact: This research addresses a critical vulnerability in AI deployment by enabling trustworthy LLM use in high-stakes domains where factual accuracy is essential, without requiring cloud connectivity or substantial computing resources.

ShED-HD: A Shannon Entropy Distribution Framework for Lightweight Hallucination Detection on Edge Devices