Aligning LLMs to Be Robust Against Prompt Injection
| Autor: | Chen, Sizhe, Zharmagambetov, Arman, Mahloujifar, Saeed, Chaudhuri, Kamalika, Guo, Chuan |
|---|---|
| Rok vydání: | 2024 |
| Předmět: | |
| Druh dokumentu: | Working Paper |
| Popis: | Large language models (LLMs) are becoming increasingly prevalent in modern software systems, interfacing between the user and the internet to assist with tasks that require advanced language understanding. To accomplish these tasks, the LLM often uses external data sources such as user documents, web retrieval, results from API calls, etc. This opens up new avenues for attackers to manipulate the LLM via prompt injection. Adversarial prompts can be carefully crafted and injected into external data sources to override the user's intended instruction and instead execute a malicious instruction. Prompt injection attacks constitute a major threat to LLM security, making the design and implementation of practical countermeasures of paramount importance. To this end, we show that alignment can be a powerful tool to make LLMs more robust against prompt injection. Our method -- SecAlign -- first builds an alignment dataset by simulating prompt injection attacks and constructing pairs of desirable and undesirable responses. Then, we apply existing alignment techniques to fine-tune the LLM to be robust against these simulated attacks. Our experiments show that SecAlign robustifies the LLM substantially with a negligible hurt on model utility. Moreover, SecAlign's protection generalizes to strong attacks unseen in training. Specifically, the success rate of state-of-the-art GCG-based prompt injections drops from 56% to 2% in Mistral-7B after our alignment process. Our code is released at https://github.com/facebookresearch/SecAlign Comment: Key words: prompt injection defense, LLM security, LLM-integrated applications. Alignment training makes LLMs robust against even the strongest prompt injection attacks |
| Databáze: | arXiv |
| Externí odkaz: |
|