beingevil.archive

curated evaluation dataset
behavior classification archive
ai-aligned validation layer

A dataset for testing the detection of “evil.”

Labels are not absolute.
Patterns are interpreted, not assigned.

...