AgentPhishing: A Benchmark of LLM Email Agents Under Adaptive Attack

No. 26 of 245 · arXiv · Justin Tan, Miao Li, Ding Zhao · published 23 Oct 2026, 04:00 UTC · 53 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

AgentPhishing is a benchmark for testing whether LLM email agents detect phishing and whether they comply with malicious requests in messages, including emails optimized against each agent. Its PhishOpt framework refines email-generation skills using agent feedback and derives a universal skill from persuasion analysis.

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

0.71total under rubric v2.0Contestedjudges differ by 0.41 on substantiated

Significance0.66
Reader interest0.77
Audience breadth0.67
Practical relevance0.70
Substantiated0.77
Technical depth0.86
New information0.90
Counts against: Promotional0.04
Counts against: Overclaim0.21

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	0.66	0.55	0.78	0.66	0.23	+0.25	+0.165
Reader interest	0.70	0.70	0.92	0.77	0.22	+0.20	+0.155
Audience breadth	0.68	0.60	0.72	0.67	0.12	+0.10	+0.067
Practical relevance	0.71	0.60	0.80	0.70	0.20	+0.15	+0.106
Substantiated	0.96	0.55	0.80	0.77	0.41 , above the contested threshold	+0.10	+0.077
Technical depth	0.94	0.70	0.93	0.86	0.24	+0.10	+0.086
New information	0.92	0.80	0.98	0.90	0.18	+0.10	+0.090
Promotional	0.05	0.05	0.01	0.04	0.04	−0.15	−0.006
Overclaim	0.19	0.35	0.08	0.21	0.27	−0.15	−0.031
Sum of contributions							+0.709
Claim-check penalty							0.000
Total, as published							0.709

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, 08:53 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.