Unlearning in LLMs leaves detectable 'fingerprints' that let a simple classifier distinguish an unlearned model from its original, even on unrelated prompts.
forget” prompt drawn from the WMDP evaluation set, which tests the model’s ability to omit specific target knowledge, and (2) a “forget- irrelevant
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
dataset 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
dataset 1polarities
use dataset 1representative citing papers
citing papers explorer
-
Unlearning Isn't Invisible: Detecting Unlearning Traces in LLMs from Model Outputs
Unlearning in LLMs leaves detectable 'fingerprints' that let a simple classifier distinguish an unlearned model from its original, even on unrelated prompts.