SoK: On the Offensive Potential of AI
This website is the complementary online tool to our paper "SoK: On the offensive potential of AI." It includes a list of academic papers (technical and non-technical) and InfoSec briefings (from BlackHat and DefCon), scrutinized based on the offensive AI checklist we developed in our paper. The checklist is a lifelong classification of the key-factors of technologies related to offensive AI, allowing us to objectively examine and compare existing and novel use-cases of offensive AI according to a clearly defined set of criteria. Further details will be provided after the paper has been published.
Below, we provide three lists: (i) for the technical papers, (ii) for the non-technical papers, and (iii) for the InfoSec briefings. The key difference between technical and non-technical papers is that technical papers must demonstrate a practical implementation of the AI model. For the Specific Offensive AI (OAI) use case we highlight papers/briefings mapped to MITRE ATT&CK in green, and papers/briefings mapped to categories unrelated to MITRE ATT&CK in blue.
If you spot any mistakes, or missing papers, feel free to contact us (anonymous for now, but we will add credentials after having published the paper), or just go ahead and submit an entry!
