| title | Quickest Detection of Hallucination Onset: Delay Bounds and Learned CUSUM Statistics |
|---|---|
| authors | Igor Itkin |
| date | 2026-06-10 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2606.12476 |
| upvotes | 1 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | Token-level hallucination detectors are evaluated as classifiers, by AUC over all tokens, yet a streaming monitor is judged by its reaction time: the number of tokens that pass between the onset of a hallucination and the alarm. We formulate hallucination onset detection as a quickest change detection problem. A first-order Markov model of the latent faithful/hallucinated state, validated on RAGTruth, places the task inside classical change-point theory and yields Lorden's lower bound on detection delay: about 1.3 tokens at a false-alarm rate of 0.01. We then show that a causal recurrent labeler acts as a CUSUM with a learned increment; at a matched false-alarm rate it detects in 11-13 tokens, against 31 for a linear per-token baseline, and a controlled decomposition attributes most of this advantage to a better per-token score rather than to temporal accumulation. An information-rate optimality theorem of Donsker-Varadhan type explains the remaining order-of-magnitude gap: the learned score realizes only 1/4.5 of the divergence the features carry, a deficit that recalibration cannot remove, with the remainder a finite-horizon effect. Classification metrics conceal this delay structure; sequential analysis makes it measurable |
cite
@misc{itkin2026quickest,
title={Quickest Detection of Hallucination Onset: Delay Bounds and Learned CUSUM Statistics},
author={Igor Itkin},
year={2026},
eprint={2606.12476},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2606.12476}
}