Applying Large Language Models to Power Systems: Potential Security Threats

Jiaqi Ruan,Gaoqi Liang,Huan Zhao,Guolong Liu,Xianzhuo Sun,Jing Qiu,Zhao Xu,Fushuan Wen,Zhao Yang Dong

2024-01-24

Abstract:Applying large language models (LLMs) to modern power systems presents a promising avenue for enhancing decision-making and operational efficiency. However, this action may also incur potential security threats, which have not been fully recognized so far. To this end, this article analyzes potential threats incurred by applying LLMs to power systems, emphasizing the need for urgent research and development of countermeasures.

Artificial Intelligence,Human-Computer Interaction,Systems and Control

What problem does this paper attempt to address?

The paper primarily explores the application of large language models (LLMs) in modern power systems and the potential security threats they may pose.

### Problems the Paper Attempts to Address

1. **Enhancing Decision-Making Efficiency and Operational Effectiveness**: With the widespread adoption of renewable energy and the integration of various entities within power systems, the complexity and openness of power systems are continuously increasing. This presents significant challenges for power system operators. To address these challenges, researchers have introduced large language models to enhance the decision-making process and improve operational efficiency.

2. **Analyzing Potential Security Threats**: Although the application of large language models in power systems holds great promise, it may also bring a series of security issues that have not yet received sufficient attention. Therefore, the paper aims to deeply analyze these potential security threats and emphasize the importance of urgent research and development of related countermeasures.

### Specific Security Threats Include:

1. **Data Privacy Infringement**: The application of large language models may make sensitive information more accessible to attackers, thereby posing a risk of data privacy breaches.

2. **Model Performance Degradation**: Malicious tampering with training data or directly modifying model parameters may lead to inaccurate model outputs, affecting the stable operation of power systems.

3. **Semantic Discrepancy Attacks (SDAs)**: Attackers may manipulate inputs or outputs, causing decision-makers to receive incorrect information, thereby impacting the reliability and efficiency of the entire power system.

4. **Denial of Service (DoS) Attacks**: By sending a large number of requests or particularly complex queries to large language models, attackers can render the models inoperative, thereby affecting the stability and security of power systems.

In summary, the core of the paper lies in revealing the potential of large language models in power systems while also pointing out the accompanying security risks. It provides corresponding suggestions to promote the secure development of future power systems.