Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attacks

Recent advances in instruction-following large language models (LLMs) have led to dramatic improvements in a range of NLP tasks. Unfortunately, we find that the same improved capabilities amplify the dual-use risks for malicious purposes of these models. Dual-use is difficult to prevent as instruction-following capabilities now enable standard attacks from computer security. The capabilities of these instruction-following LLMs provide strong economic incentives for dual-use by malicious actors. In particular, we show that instruction-following LLMs can produce targeted malicious content, including hate speech and scams, bypassing in-the-wild defenses implemented by LLM API vendors. Our analysis shows that this content can be generated economically and at cost of $125-500 \times$ cheaper than human effort alone. Together, our findings suggest that LLMs will increasingly attract more sophisticated adversaries and attacks, and addressing these attacks may require new approaches to mitigations.

1. Introduction

Large language models (LLMs) have recently improved dramatically in text generation. This improvement is driven in large part by scale and the ability to be instruction following [1], [2], [3], [4], [5], [6], [7]. As with most technologies, LLMs have a potential for dual-use, where their language generation capabilities are used for malicious or nefarious ends. For example, text generation models have already been used to produce hateful text [8].