This is the repository of the paper "SoK: On the Offensive Potential of AI", accepted to the 3rd IEEE Conference on Secure and Trustworthy MachineLearning (SaTML'25).

If you use any of our resources, you are kindly invited to cite our paper:

@inproceedings{schroeer2025sok,
    title={{SoK: On the Offensive Potential of AI}},
    author={Schröer, Saskia Laura and Apruzzese, Giovanni and Soheil, Human and Laskov, Pavel and Anderson, Hyrum S. and Bernroider, Edward W. N. and Fass, Aurore and Nassi, Ben and Rimmer, Vera and Roli, Fabio and Salam, Samer and Shen, Ashley and Sunyaev, Ali and Wadhwa-Brown, Tim and Wagner, Isabel and Wang, Gang},
    booktitle={Proceedings of the IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)},
    year={2025},
    publisher={IEEE}
}

The repository contains the following files:

• Two jupyter notebooks (analysis-BERTopic.ipynb and analysis-keyBERT.ipynb) containing the code for the NLP-related analyses.
• A demonstrative video (demo.mp4) of our website (reachable at: https://sok-offensive-ai.github.io/)
• The surveys we used to collect the opinions of laypeople (laypeople_survey.pdf) and for the initial opinions of the experts (expert_survey.pdf)
• The instructions that the experts had to follow when providing their statements (statement_instructions.pdf)

If you have any inquiry about our research (e.g., about the code) feel free to contact the first author, Dr. Saskia Laura Schröer (saskia.schroeer@uni.li)