---
id: 20261007-T0-19
title: "Proxy Confidence: Auditing Black-Box LLM Agents with a Surrogate's Log-Probabilities"
url: https://ai.daily.yangsir.net/daily/20261007-T0-19
issue_date: 2026-10-07
publish_date: 2026-10-06T04:00:00.000Z
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2610.03894
---

# Proxy Confidence: Auditing Black-Box LLM Agents with a Surrogate's Log-Probabilities

arXiv:2610.03894v1 Announce Type: new Abstract: A deployed LLM agent emits tool calls, queries, and code that can be silently wrong -- by the time the error surfaces, the action has run. Frontier chat APIs hide the model's token probabilities; the agent's stated confidence barely beats chance on the

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2610.03894)

**详情页**：https://ai.daily.yangsir.net/daily/20261007-T0-19

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*