Integrating Blockchain with LLMs: Towards a Secure and Safe Technology
Faculty Mentor Information
Dr. Gaby Dagher (Mentor), Boise State University
Presentation Date
7-2024
Abstract
Large language models (LLMs) are an emerging technology with the potential to drastically affect how we interact with computing systems. They can produce code, write essays, and be used in agentic capacities. However, various vulnerabilities within these systems have been discovered and current solutions cannot entirely placate the wide array of existing threats. For instance, some vulnerabilities allow the system to respond with misinformation and render the system dysfunctional when attacked. In lieu of these problems, we offer a comprehensive investigation into the rapidly developing field of blockchain for LLMs (BC4LLMs). In particular, we analyze the field concerning blockchain-based security and safety mechanisms. In our survey of BC4LLMs we propose novel definitions for security and safety regarding LLMs and utilize these to further contextualize research efforts. We categorize and elicit prominent areas of interest to those working within this field as well as aim to introduce a common consensus on secure and safe LLMs. Limitations and future research directions are discussed; special emphasis is placed on blockchain’s beneficial integration with this potentially disruptive technology.
Integrating Blockchain with LLMs: Towards a Secure and Safe Technology
Large language models (LLMs) are an emerging technology with the potential to drastically affect how we interact with computing systems. They can produce code, write essays, and be used in agentic capacities. However, various vulnerabilities within these systems have been discovered and current solutions cannot entirely placate the wide array of existing threats. For instance, some vulnerabilities allow the system to respond with misinformation and render the system dysfunctional when attacked. In lieu of these problems, we offer a comprehensive investigation into the rapidly developing field of blockchain for LLMs (BC4LLMs). In particular, we analyze the field concerning blockchain-based security and safety mechanisms. In our survey of BC4LLMs we propose novel definitions for security and safety regarding LLMs and utilize these to further contextualize research efforts. We categorize and elicit prominent areas of interest to those working within this field as well as aim to introduce a common consensus on secure and safe LLMs. Limitations and future research directions are discussed; special emphasis is placed on blockchain’s beneficial integration with this potentially disruptive technology.