Exploiting Novel GPT-4 APIs
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909278638440448 |
|---|---|
| author | Pelrine, Kellin Taufeeque, Mohammad Zając, Michał McLean, Euan Gleave, Adam |
| author_facet | Pelrine, Kellin Taufeeque, Mohammad Zając, Michał McLean, Euan Gleave, Adam |
| contents | Language model attacks typically assume one of two extreme threat models: full white-box access to model weights, or black-box access limited to a text generation API. However, real-world APIs are often more flexible than just text generation: these APIs expose "gray-box" access leading to new threat vectors. To explore this, we red-team three new functionalities exposed in the GPT-4 APIs: fine-tuning, function calling and knowledge retrieval. We find that fine-tuning a model on as few as 15 harmful examples or 100 benign examples can remove core safeguards from GPT-4, enabling a range of harmful outputs. Furthermore, we find that GPT-4 Assistants readily divulge the function call schema and can be made to execute arbitrary function calls. Finally, we find that knowledge retrieval can be hijacked by injecting instructions into retrieval documents. These vulnerabilities highlight that any additions to the functionality exposed by an API can create new vulnerabilities. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2312_14302 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Exploiting Novel GPT-4 APIs Pelrine, Kellin Taufeeque, Mohammad Zając, Michał McLean, Euan Gleave, Adam Cryptography and Security Artificial Intelligence Computation and Language Machine Learning I.2.7 Language model attacks typically assume one of two extreme threat models: full white-box access to model weights, or black-box access limited to a text generation API. However, real-world APIs are often more flexible than just text generation: these APIs expose "gray-box" access leading to new threat vectors. To explore this, we red-team three new functionalities exposed in the GPT-4 APIs: fine-tuning, function calling and knowledge retrieval. We find that fine-tuning a model on as few as 15 harmful examples or 100 benign examples can remove core safeguards from GPT-4, enabling a range of harmful outputs. Furthermore, we find that GPT-4 Assistants readily divulge the function call schema and can be made to execute arbitrary function calls. Finally, we find that knowledge retrieval can be hijacked by injecting instructions into retrieval documents. These vulnerabilities highlight that any additions to the functionality exposed by an API can create new vulnerabilities. |
| title | Exploiting Novel GPT-4 APIs |
| topic | Cryptography and Security Artificial Intelligence Computation and Language Machine Learning I.2.7 |
| url | https://arxiv.org/abs/2312.14302 |