The Public Repository used by the Frankfurt AI safety (FAIS) group for their AI safety replication project.
This repository will be used to host the code used for replicating AI models, our safety meassurement metrics, and resources used.
Subject to change
| Kick-off | Define Ai Safety Metrics | Reproduce Model with regard to Safety Metrics | Write Report | Finishing Touches and Publishing |
|---|---|---|---|---|
| 28. May 2026 | 11. June 2026 | 1. October 2026 | 12. November 2026 | 10. December 2026 |
| Which paper/model are we replicating? | Define Metrics | Reproduce model | Write the Report | Publish the Report |
| Quantify needed resources | Ensure no Safetywashing | Check model against metrics | Produce Illustrative Graphs | |
| Define Milestones | Incorporate EU AI Act |
Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress?
Stable Diffusion Safety Checker
AISafetyLab: A Comprehensive Framework for AI Safety Evaluation and Improvement