Skip to content

Latest commit

 

History

History
20 lines (14 loc) · 1.02 KB

README.md

File metadata and controls

20 lines (14 loc) · 1.02 KB

Investigating Prompt-Induced Bias in GPT-3.5 Turbo Responses within Digital Forensics

This repository contains experiments that we performed on the Honeynet challenge 3 "Banking Troubles" dataset and their results.

Experiments were performed on the questions 1, 2, 7 and 9 of the original challenge. Questions were prompted to the GPT-3.5 Turbo model using the OpenAI API.

In each case the original question was kept as a baseline and multiple biased questions were constructed to evaluate the model's susceptibility to bias in the prompts.

For each question additional context was produced by running volatility against the memory dump provided in the challenge.

Datasets for different Honeynet challenges can be found here: https://github.com/honeynet/forensic_challenges/tree/master

To execute experiments locally, make sure to have a OPENAI_API_KEY environment variable set with your OpenAI API key.