An official website of the United States government
Large Language Models (LLMs), such as ChatGPT and Bard, have revolutionized natural language understanding and generation. They possess deep language comprehension, human-like text generation capabilities, contextual awareness, and robust problem-solving skills, making them invaluable in various domains (e.g., search engines, customer support, translation). In the meantime, LLMs have also gained traction in the security community, revealing security vulnerabilities and showcasing their potential in security-related tasks. This paper explores the intersection of LLMs with security and privacy. Specifically, we investigate how LLMs positively impact security and privacy, potential risks and threats associated with their use, and inherent vulnerabilities within LLMs. Through a comprehensive literature review, the paper categorizes the papers into “The Good” (beneficial LLM applications), “The Bad” (offensive applications), and “The Ugly” (vulnerabilities of LLMs and their defenses). We have some interesting findings. For example, LLMs have proven to enhance code security (code vulnerability detection) and data privacy (data confidentiality protection), outperforming traditional methods. However, they can also be harnessed for various attacks (particularly user-level attacks) due to their human-like reasoning abilities. We have identified areas that require further research efforts. For example, Research on model and parameter extraction attacks is limited and often theoretical, hindered by LLM parameter scale and confidentiality. Safe instruction tuning, a recent development, requires more exploration. We hope that our work can shed light on the LLMs’ potential to both bolster and jeopardize cybersecurity.
more »
« less
This content will become publicly available on July 20, 2027
Measuring Reproducibility in LLM Research: A Rubric and Cross-Topic Corpus Study
Large language model (LLM) research now spans all computer research areas, yet the practical reproducibility of that literature remains uneven. Releasing code is no longer enough to ensure reproducibility because LLM-related papers may rely on gated model weights, mutable application programming interfaces (APIs), hidden system prompts, unstable dependency stacks, expensive accelerator resources, or evaluation backends that drift after publication. We present a cross-topic corpus study of LLM papers covering six categories: foundation models, alignment, retrieval-augmented generation (RAG) and agent systems, code generation, security/privacy, and evaluation/domain systems. Each paper is scored on five axes: model availability, software packaging, data accessibility, hardware approachability, and evaluation durability. The resulting scores estimate rerun feasibility from the public artifact without end-to-end reproduction. We report three findings, (1) software and benchmark artifacts are often more durable than model access: many papers remain inspectable or partially rerunnable even when the original model cannot be reused directly; (2) code generation and evaluation/domain papers are the most reproducible in our sample because they frequently pair LLM outputs with executable checks; (3) foundation-model papers remain constrained by compute, while security/privacy papers are often cheap to rerun but fragile because they depend on mutable APIs and interfaces. We provide empirical evidence about how models, software, hardware, and evaluation drift shape the practical reproducibility of LLM research.
more »
« less
- Award ID(s):
- 2453331
- PAR ID:
- 10700427
- Publisher / Repository:
- Association for Computing Machinery
- Date Published:
- ISBN:
- 979-8-4007-2778-8
- Subject(s) / Keyword(s):
- large language models reproducibility replicability research artifacts software dependencies empirical studies natural language processing empirical software validation software and application security
- Format(s):
- Medium: X
- Location:
- Delft, Netherlands
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Large language models (LLMs) are widely used in software development. However, the code generated by LLMs often contains vulnerabilities. Several secure code generation methods have been proposed to address this issue, but their current evaluation schemes leave several concerns unaddressed. Specifically, most existing studies evaluate security and functional correctness separately, using different datasets. That is, they assess vulnerabilities using securityrelated code datasets while validating functionality with general code datasets. In addition, prior research primarily relies on a single static analyzer, CodeQL, to detect vulnerabilities in generated code, which limits the scope of security evaluation. In this work, we conduct a comprehensive study to systematically assess the improvements introduced by four state-of-the-art secure code generation techniques. Specifically, we apply both security inspection and functionality validation to the same generated code and evaluate these two aspects together. We also employ three popular static analyzers and two LLMs to identify potential vulnerabilities in the generated code. Our study reveals that existing techniques often compromise the functionality of generated code to enhance security. Their overall performance remains limited when evaluating security and functionality together. In fact, many techniques even degrade the performance of the base LLM by more than 50%. Our further inspection reveals that these techniques often either remove vulnerable lines of code entirely or generate “garbage code” that is unrelated to the intended task. Moreover, the commonly used static analyzer CodeQL fails to detect several vulnerabilities, further obscuring the actual security improvements achieved by existing techniques. Our study serves as a guideline for a more rigorous and comprehensive evaluation of secure code generation performance in future work.more » « less
-
Large language models (LLMs) are widely used in software development. However, the code generated by LLMs often contains vulnerabilities. Several secure code generation methods have been proposed to address this issue, but their current evaluation schemes leave several concerns unaddressed. Specifically, most existing studies evaluate security and functional correctness separately, using different datasets. That is, they assess vulnerabilities using securityrelated code datasets while validating functionality with general code datasets. In addition, prior research primarily relies on a single static analyzer, CodeQL, to detect vulnerabilities in generated code, which limits the scope of security evaluation. In this work, we conduct a comprehensive study to systematically assess the improvements introduced by four state-of-the-art secure code generation techniques. Specifically, we apply both security inspection and functionality validation to the same generated code and evaluate these two aspects together. We also employ three popular static analyzers and two LLMs to identify potential vulnerabilities in the generated code. Our study reveals that existing techniques often compromise the functionality of generated code to enhance security. Their overall performance remains limited when evaluating security and functionality together. In fact, many techniques even degrade the performance of the base LLM by more than 50%. Our further inspection reveals that these techniques often either remove vulnerable lines of code entirely or generate “garbage code” that is unrelated to the intended task. Moreover, the commonly used static analyzer CodeQL fails to detect several vulnerabilities, further obscuring the actual security improvements achieved by existing techniques. Our study serves as a guideline for a more rigorous and comprehensive evaluation of secure code generation performance in future work.more » « less
-
This paper presents LLM4SecHW, a novel framework for hardware debugging that leverages domain-specific Large Language Model (LLM). Despite the success of LLMs in automating various software development tasks, their application in the hardware security domain has been limited due to the constraints of commercial LLMs and the scarcity of domain-specific data. To address these challenges, we propose a unique approach to compile a dataset of open-source hardware design defects and their remediation steps, utilizing version control data. This dataset provides a substantial foundation for training machine learning models for hardware. LLM4SecHW employs fine-tuning of medium-sized LLMs based on this dataset, enabling the identification and rectification of bugs in hardware designs. This pioneering approach offers a reference workflow for the application of fine-tuning domain-specific LLMs in other research areas. We evaluate the performance of our proposed system on various open-source hardware designs, demonstrating its efficacy in accurately identifying and correcting defects. Our work brings a new perspective on automating the quality control process in hardware design.more » « less
-
Large Language Models (LLMs) have extensive ability to produce promising output. Nowadays, people are increasingly relying on them due to easy accessibility, rapid and outstanding outcomes. However, the use of these results without appropriate scrutiny poses serious security risks, particularly when they are integrated with other software, APIs, or plugins. This is because the LLM outputs are highly dependent on the prompts they receive. Therefore, it is essential to carefully clean these outputs before using them in additional software environments. This paper is designed to teach students about the potential dangers of contaminated LLM output within the context of web development through prelab, handson, and postlab experiences. Hands-on lab provides practical guidance on how to handle LLM vulnerabilities to make applications safe with some real-world examples in Python. This approach aims to provide students with a deeper understanding of the precautions necessary to ensure software against the vulnerabilities introduced by LLM output.more » « less
