Year: 2026
Type: preprint
Abstract: Recent work has shown that a teacher model can transfer a bias to a student through a dataset from which every explicit reference to that bias has been filtered out, and that none of the tested data-level defenses reliably removes or detects such a bias, even when the defender knows what to look for. To shed light on the hidden traces these biases leave, we embed a dataset's completions with Sentence-BERT, subtract the embeddings of clean reference completions, and compare the result to an open vocabulary of candidate topics. This simple signature identifies the topic of the bias with a Matthews correlation coefficient of 0.83 when the attacker's teacher model is known, and 0.46 when it is not. We also observe that different teacher models appear to express the same bias through different vocabulary.
Source: arXiv (Cornell University)
Authors: Jürß, Jonas, Liò, Pietro
Institution: University of Cambridge
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7213346556
Cites:
Cited by:
Related to: