---
格式版本: 2
标题: "The Acknowledgment Point Is the System: Durable Policy-Decision Receipts for AI Audit Evidence"
原文链接: "https://arxiv.org/abs/2608.17176"
发布日期: "2026-08-17"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Mon, 17 Aug 2026 22:35:07 UTC (63 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T15:40:32+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T15:33:55+08:00"
入库时间: "2026-08-20T07:40:32.847Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=latency&searchtype=all"
匹配关键词:
  - "latency"
  - "throughput"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 5
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "论文主题为AI审计证据与密码学安全，仅摘要提及latency，与超节点/AI Rack/机柜级AI基础设施完全无关，属于明显无关内容。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T15:45:25+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 0
AI摘要: "该研究重建了AI审计原型RuntimeGuard-AI，将每次确定性政策决策绑定到确切策略源并在调用者选择的同步边界提交最小隐私记录，返回Ed25519签名回执以确认持久化完成。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:24:47.156Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.17176"
---

## Computer Science > Cryptography and Security

## Title:The Acknowledgment Point Is the System: Durable Policy-Decision Receipts for AI Audit Evidence

Authors:[Neeraj Kumar Singh Beshane](https://arxiv.org/search/cs?searchtype=author&query=Beshane,+N+K+S)

[View PDF](https://arxiv.org/pdf/2608.17176) [HTML (experimental)](https://arxiv.org/html/2608.17176v1)

> Abstract:An AI audit record is useful only if its durability and trust boundary are explicit. Returning a guarded decision before any durable write minimizes latency, but it cannot guarantee that evidence survives an immediate crash. We rebuild RuntimeGuard-AI around this constraint. The resulting research prototype binds each deterministic policy decision to the exact policy source, commits a privacy-minimizing record at a caller-selected synchronization boundary, and returns an Ed25519-signed receipt that states whether that boundary completed. After restart, the engine validates framed records, manifests, shard placement, sequence continuity, and replay identity. A separate attestation path groups committed records into chained, signed Merkle epochs that an auditor verifies with an externally obtained key. On an Apple M4 Pro at four worker threads and 2,048-byte prompts, buffered signed evidence reaches 27,193 requests/s with 141.9 microseconds median latency. Per-record data and full synchronization reduce throughput to approximately 242 requests/s and raise median latency to 16.0 ms. Sealing a 100,000-record signed epoch takes 97.0 ms. The result is a measured durability-latency trade-off, not a "free" asynchronous audit path. The prototype does not prove model execution, prevent a compromised signer from forking history, or establish legal conformity.

| Comments: |  |
| --- | --- |
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) |
| Cite as: | [arXiv:2608.17176](https://arxiv.org/abs/2608.17176) \[cs.CR\] |
|  | (or [arXiv:2608.17176v1](https://arxiv.org/abs/2608.17176v1) \[cs.CR\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.17176](https://doi.org/10.48550/arXiv.2608.17176) |

## Submission history

From: Neeraj Kumar Singh Beshane \[[view email](https://arxiv.org/show-email/0f5f11e5/2608.17176)\]  
**\[v1\]** Mon, 17 Aug 2026 22:35:07 UTC (63 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.17176) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))