Fundamental Risks in the Current Deployment of General-Purpose AI Models: What Have We (Not) Learnt From Cybersecurity?
It highlights risks for users and developers deploying AI in critical systems, but is incremental as it builds on existing cybersecurity concerns.
The paper addresses cybersecurity challenges arising from the rapid deployment of general-purpose AI models like LLMs, which are increasingly used in autonomous applications with data access and execution capabilities, but does not report specific results or numbers.
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.