[FEATURE] New notebook for detecting python code vulnerabilities using GPT4

[mhbuehler]

Is there any interest in a notebook contribution that demonstrates a few ways of prompting GPT4 to detect insecure code snippets? The solution would include prompt templates that use techniques like few-shot learning, KNN-based few-shot learning (from this paper), and asking for a code fix (proposed by this paper) to increase prediction accuracy.

Is your feature request related to a problem? Please describe. This is a proposal for a new use case. It would help users who are looking for examples that:

• Use the OpenAI API to perform binary classification
• Apply the above prompt engineering techniques and evaluate their impacts
• Identify and/or correct software vulnerabilities

Describe the solution you'd like I've written a draft notebook based on my experiments and would be happy to submit a PR. I'm an experienced software engineer and have recently been applying AI to topics in software security.

Additional context There would be a new data file based on the published dataset from this paper which has this license. The notebook would cite all sources and include some evaluation metrics. Thanks for your consideration.

[github-actions[bot]]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[mhbuehler]

Commenting so this issue stays open. Can maintainers take a look at the notebook in https://github.com/openai/openai-cookbook/pull/1112 and let me know if it can be merged or not? Thank you.

[github-actions[bot]]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[mhbuehler]

Commenting to remove the stale label