LLM framework BadQubits detects harmful quantum circuits before execution

Researchers have developed BadQubits, a framework utilizing Large Language Models (LLMs) to detect harmful quantum circuits before execution. This system analyzes OpenQASM 2.0 circuits to identify physical-execution-layer threats, which are difficult to detect during runtime due to measurement irreversibility and simulation costs. A fine-tuned Qwen Coder 2.5 7B model achieved 92.67% classification accuracy and 96.1% recall for harmful circuits, outperforming a bag-of-gates CNN in identifying specific threat features like SWAP density and measurement timing. AI

IMPACT This research demonstrates a novel application of LLMs in identifying security threats within quantum computing, potentially enhancing the safety and integrity of quantum circuit execution.

RANK_REASON Research paper detailing a novel framework for detecting harmful quantum circuits using LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• BadQubits
• CNN
• LoRA+
• MQTBench
• OpenQASM 2.0
• Qwen Coder 2.5 7B

• paper
• safety
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →