Beyond Causal Masks: Auditing and Localizing Hidden Future-Token Leakage in Adaptive Sequence Models

No. 91 of 245 · arXiv · Huiting Liu, Jinsong Wu · published 23 Oct 2026, 04:00 UTC · 40 min read

AI & machine learning
Paper

3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked

A paper in Neurocomputing examines hidden future-token leakage, where an autoregressive model reads information from later tokens through a non-causal auxiliary path even when its attention mask is correctly causal, producing misleadingly strong teacher-forced results.

Read the original at arxiv.org. The judges saw only its text, with the site, the author and link domains removed; they are shown again after ranking.

Score

0.66total under rubric v2.0

Significance0.57
Reader interest0.69
Audience breadth0.56
Practical relevance0.74
Substantiated0.80
Technical depth0.90
New information0.85
Counts against: Promotional0.04
Counts against: Overclaim0.22

Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).

Question	Judge 1	Judge 2	Judge 3	Mean	Spread	Weight	Contribution (weight times mean)
Significance	0.62	0.40	0.68	0.57	0.28	+0.25	+0.142
Reader interest	0.58	0.60	0.88	0.69	0.30	+0.20	+0.137
Audience breadth	0.55	0.50	0.63	0.56	0.13	+0.10	+0.056
Practical relevance	0.74	0.70	0.78	0.74	0.08	+0.15	+0.111
Substantiated	0.94	0.70	0.76	0.80	0.24	+0.10	+0.080
Technical depth	0.99	0.80	0.92	0.90	0.19	+0.10	+0.090
New information	0.89	0.70	0.97	0.85	0.27	+0.10	+0.085
Promotional	0.05	0.05	0.02	0.04	0.03	−0.15	−0.006
Overclaim	0.23	0.30	0.12	0.22	0.18	−0.15	−0.033
Sum of contributions							+0.663
Claim-check penalty							0.000
Total, as published							0.663

Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.

Also covered by

No other coverage was clustered with this story.

Elsewhere

Elsewhere: HN —

Recorded for context only, checked 24 Oct 2026, 08:54 UTC. It played no part in the ranking.

Panel for this edition

Four AI agents ranked this edition: three judges scored every story blind to its site and author, and a fourth compared the top stories’ claims with the sources they cite. Summaries are written by AI.

Model versions are pinned for every edition and listed in its JSON. This one was scored under rubric v2.0; How ranking works describes the current rubric, v2.2.