A key lesson of recent deep learning successes is that as we scale neural networks, they get better, and sometimes in game-changing ways. This repo provides two applications which demonstrate how GPT-3 opens new vistas for cybersecurity.
There are two use cases for spam detection and command analysis. As the code in this repo uses OpenAI API, set the OPENAI_API_KEY enviroment variable as your api key. Refer to the OpenAI documentation in https://beta.openai.com/docs/introduction.
Spam detector demonstrates how to identify spam messages using GPT-3 few-shot learning or fine-tuning.
Change directory to spam_detector folder and follow the instructions.
Command analyzer shows how to analyzer complex command lines using GPT-3 few-shot learning.
Change directory to command_analyzer folder and follow the instructions.
Questions, ideas, feedback appreciated, please email younghoo.lee@sophos.com
@misc{Lee2022, author = {Lee, Younghoo}, title = {GPT-3 for Cybersecurity}, year = {2022}, publisher = {GitHub}, journal = {GitHub repository}, howpublished = {\url{https://github.com/sophos-ai/gpt3-cybersecurity/}} }