# Preprint measures evidence replication risk in AI-agent reports

_Published Thursday, September 3, 2026 at 7:06 AM EDT · Science, AI · Latest · Tier 2 — Notable_

A preprint formalizes an "epistemic Sybil" problem in multi-agent AI: additional reports can stem from the same evidence rather than provide independent corroboration. In more than 20,000 controlled LLM-agent calls on synthetic evidentiary documents, the authors report naive posterior coverage fell from 0.940 to 0.263 when report multiplicity rose from one to 32 with one evidence root. They also report that a correlated-extraction aggregator restored calibration.

## Sources

- [cs.AI updates on arXiv.org](https://arxiv.org/abs/2609.01873)

---
Canonical: https://techandbusiness.org/newswire/VUVJdh-UKmuTSaO2KHPM3K
Published: 2026-09-03T11:06:47.116Z
Story chronology: 2026-09-03T04:00:00.000Z
Retrieved: 2026-10-18T18:16:06.971Z
Publisher: Tech & Business (techandbusiness.org)