Independent researcher. Working on falsifiable architectural approaches to AI alignment (non-dual optimization, mesa-optimization prevention).

doi.org/10.5281/zenodo.21916934

doi.org/10.5281/zenodo.21913550