Large Language Model (LLM) is a significant breakthrough in artificial intelligence (AI) and holds considerable potential for application within smart grids. However, as demonstrated in previous literature, AI technologies are susceptible to various types of attacks. It is crucial to investigate and evaluate the risks associated with LLMs before deploying them in critical infrastructure like smart grids. In this paper, we systematically evaluate the vulnerabilities of LLMs and identify two major types of attacks relevant to smart grid LLM applications, along with presenting the corresponding threat models. We then validate these attacks using popular LLMs, utilizing real smart grid data. Our validation demonstrates that attackers are capable of injecting bad data and retrieving domain knowledge from LLMs employed in smart grid scenarios.

翻译：大语言模型（LLM）是人工智能（AI）领域的一项重大突破，在智能电网中具有广阔的应用潜力。然而，正如先前文献所示，AI技术易受各类攻击。在智能电网等关键基础设施中部署LLM之前，必须对其相关风险进行调查与评估。本文系统评估了LLM的脆弱性，识别出与智能电网LLM应用相关的两类主要攻击类型，并提出了相应的威胁模型。随后，我们利用实际智能电网数据，在主流LLM上验证了这些攻击。验证结果表明，攻击者能够向应用于智能电网场景的LLM中注入恶意数据并窃取领域知识。