PARALLAX: Separating Genuine Progress from Benchmark Artifacts in Hallucination Detection

Khizar Hussain, Murat Kantarcioglu · 2026

Read PARALLAX: Separating Genuine Progress from Benchmark Artifacts in Hallucination Detection by Khizar Hussain, Murat Kantarcioglu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-17028.txt.

Description

Hussain and Kantarcioglu show much progress on hallucination detection rests on benchmark artifacts, not real capability. Four of six corpora embed the answer in the input prompt, letting a naive text-similarity baseline reach near-perfect scores with no model internals. Their large evaluation spans twenty-two detection methods, twelve open-source models across six families, and six corpora, plus DRIFT, a supervised probe over hidden-state transitions. Most established baselines score near chance under controlled conditions; the exceptions are SAPLMA and DRIFT, both supervised probes on upper hidden states.

Catalog record

CREATORKhizar Hussain, Murat Kantarcioglu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE61051
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.17028

Browse the full library · About the Tower of Babel