Fundamental Risks in the Current Deployment of General-Purpose AI Models: What Have We (Not) Learnt From Cybersecurity?

TL;DR

This paper examines the cybersecurity risks associated with deploying general-purpose AI models like LLMs, highlighting challenges and lessons learned from cybersecurity to inform safer deployment practices.

Contribution

It analyzes current risks in deploying G-P AI models and draws lessons from cybersecurity to improve safety and robustness.

Findings

Identification of key cybersecurity vulnerabilities in G-P AI deployment

Evaluation of existing risks and mitigation strategies

Outline of future research directions for safer AI deployment

Abstract

General Purpose AI - such as Large Language Models (LLMs) - have seen rapid deployment in a wide range of use cases. Most surprisingly, they have have made their way from plain language models, to chat-bots, all the way to an almost ``operating system''-like status that can control decisions and logic of an application. Tool-use, Microsoft co-pilot/office integration, and OpenAIs Altera are just a few examples of increased autonomy, data access, and execution capabilities. These methods come with a range of cybersecurity challenges. We highlight some of the work we have done in terms of evaluation as well as outline future opportunities and challenges.