Our work on the vulnerability of concept erasure methods is out on is accepted to ICLR 2024. Check out our project website!