Field building
Community
Adversarial interpretability has no dedicated institution — so part of the work is building the field itself.
Alongside the research, we teach: an open curriculum, public talks, and hands-on workshops that take a cohort from linear probes to detection tooling that holds up when models are actively non-cooperative — with a running log of what we work through, week by week.
Curriculum 3 modules
A syllabus for the field
The course we wish existed — from activations and probes to detection under adversarial pressure.
Explore Talks 2 talks
Talks & lectures
Public talks and guest lectures introducing adversarial interpretability to new audiences.
Explore Workshops 2 workshops
Teaching the cohort
Hands-on workshops where cohorts build their first working interpretability tooling.
Explore Research log 3 entries
What has happened so far
A running log of what the community works through, week by week.
Explore