Agent Mosaic: Recognizing Confidentiality Leaks in Agentic LLM Systems

No. 68 of 245 · arXiv · Daniel Ruiz, Luca Luceri, Emilio Ferrara · published 23 Oct 2026, 04:00 UTC · first published 21 Oct 2026 · 27 min read

AI & machine learning
Security
Paper

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

The paper "Agent Mosaic" studies whether LLM agents leak confidential conclusions even when every source document they read is individually accessible to the user. Using 30 synthetic cybersecurity, medical, and military examples, the authors measure how often agents disclose a designated higher-classified composite under benign, explicit, and adversarial queries, in…

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

0.67total under rubric v2.0

Significance0.64
Reader interest0.72
Audience breadth0.59
Practical relevance0.67
Substantiated0.81
Technical depth0.81
New information0.86
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	0.69	0.50	0.72	0.64	0.22	+0.25	+0.159
Reader interest	0.69	0.60	0.88	0.72	0.28	+0.20	+0.145
Audience breadth	0.71	0.50	0.56	0.59	0.21	+0.10	+0.059
Practical relevance	0.71	0.60	0.68	0.67	0.12	+0.15	+0.100
Substantiated	0.95	0.70	0.79	0.81	0.25	+0.10	+0.081
Technical depth	0.96	0.60	0.86	0.81	0.36	+0.10	+0.081
New information	0.89	0.70	0.98	0.86	0.28	+0.10	+0.086
Promotional	0.05	0.05	0.01	0.04	0.04	−0.15	−0.006
Overclaim	0.23	0.30	0.12	0.22	0.18	−0.15	−0.033
Sum of contributions							+0.672
Claim-check penalty							0.000
Total, as published							0.672

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.