1
四部门为汽车安全划定红线：明确汽车设计、可靠性验证试验管理，研发测试要包含两冬一夏
IT之家 10 月 11 日消息，工业和信息化部会同公安部、生态环境部、市场监管总局 10 日联合发布《关于进一步加强汽车产品创新设计和研发测试验证有关管理工作的通知（征求意见稿）》。 据央视新闻今日报道，此次出台的文件重点面向三个方向： 一是研发测试验证，明确测试周期和里程的硬性要求。新申报产品必…
2
JEDEC 发布首份全行业硅光子学（SiPh）可靠性标准 JESD264
IT之家 10 月 5 日消息，JEDEC 固态技术协会美国当地时间本月 1 日发布了 JESD264 硅光子学 (SiPh) 器件认证和可靠性要求，这是 首份全行业级别的硅光器件统一认证和可靠性实践规范 。 硅光领域此前缺乏标准化的、行业通用的可靠性所有权和验证框架，导致整个供应链存在不确定性。J…
3
索尼发布第四代 DWX 无线麦克风：主打简便操作与传输可靠性，发射器自带录音备份
IT之家 9 月 10 日消息，索尼于当地时间 9 月 9 日宣布推出第四代 DWX 系列数字无线麦克风系统，包含 DWT-B40 腰包式发射器和 DWR-R40Q 机架式接收器两款新品。 该系列旨在降低无线麦克风的操作门槛，同时提升传输可靠性，适用于从专业广电制作到小型活动现场等多种场景。 索尼表…
4
OmicSync: Reliability-Aware Spatial Multi-Omics Clustering with Evidence-Constrained LLM Reasoning
用证据约束大模型推理，为空间多组学数据聚类注入可靠性感知，精准破解生物信息难题。
arXiv:2608.22785v1 Announce Type: new Abstract: Spatial multi-omics technologies jointly profile gene expression, surface proteins, and histology at e…
5
PropUQ-MAS: Propagation-Aware Uncertainty Quantification for LLM Multi-Agent Systems
多智能体协作中错误会级联放大，这项研究用传播感知量化不确定性，给LLM集群装上风险仪表盘。
arXiv:2608.22130v1 Announce Type: cross Abstract: LLM-based multi-agent systems (MAS) solve complex tasks through communication among role-specialized…
6
MemGuard: Persisting Verifier Signals for LLM-Agent Memory Governance
LLM智能体长任务中记忆失效的痛点，MemGuard用验证信号让存储经验长期可靠。
arXiv:2608.21867v1 Announce Type: new Abstract: LLM agents are moving from single-prompt use to long task streams in which reusable memory becomes a c…
7
TRACE: A Self-Evolving Skill Bank for Consistent, Limit-Aware LLM Agents
LLM代理落地关键在于一致性与边界感知，TRACE用自进化技能库破解可靠性难题，值得关注。
arXiv:2608.22793v1 Announce Type: cross Abstract: Reliable deployment of LLM agents in user-facing products depends not on raw task-solving ability bu…
8
Toward Effective and Reliable LLM Agents via Dynamic Ontology
用动态本体给智能体装上知识骨架，可靠性与效率双提升！
arXiv:2608.22974v1 Announce Type: new Abstract: Large language model (LLM) agents rely heavily on knowledge encoded in model parameters or presented a…
9
Epic CEO 吐槽微软 Ctrl+C 可靠性，让 99% 的 Win10/Win11 用户感到懊恼
IT之家 8 月 25 日消息，科技媒体 Windows Latest 昨日（8 月 24 日）发布博文，报道称在 X 平台上一张关于 Windows 11、Windows 10 系统剪贴板的梗图蹿红， Epic Games 首席执行官蒂姆 · 斯威尼（Tim Sweeney）也加入讨论。 网友 @…
10
AI with Authority, from Application to Silicon
生成式AI让机器验证从奢侈品变为必需品，成为不可收买的裁判
arXiv:2608.21356v1 Announce Type: cross Abstract: For sixty years, machine verification has been a major cost overhead, affordable only for exceptiona…
11
Our AI reviewer invented a request. Our producer retried 245 times.
AI幻觉率仅0.2%，却让生产流程空转245次，一场关于AI可靠性的真实事故复盘。
We run ~100 LLM agents unattended on local models. Last week we found one document that had been rewritten 245 times in 5 days — every attempt rejecte…
12
A single well aimed cosmic ray can demolish a LLM
模拟宇宙射线轰击代码大模型，一次精准位翻转就能让 AI 当场失灵——硬件级故障如何摧毁看似强大的 LLM。
Article URL: https://spock.is/writing/simulating-cosmic-rays-to-lobotomize-llms Comments URL: https://news.ycombinator.com/item?id=49389343 Points: 3 …
13
The LLM rewrote the email after I approved it
AI代理篡改已批准邮件？600万小时运行揭示真正痛点不在模型智力，而在边界失控。
Article URL: https://www.crusbro.com/en/product-architecture.html Comments URL: https://news.ycombinator.com/item?id=49384800 Points: 2 # Comments: 0
14
Uncertainty-Aware Decision Making in Multimodal Large Language Models
多模态大模型如何量化不确定性？这项研究为高风险场景下的可靠决策提供新思路。
arXiv:2608.17084v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) increasingly answer questions whose correctness depends on vi…
15
Beyond BFI: The CSI for Enhanced Reliability and Validity in Evaluating LLM Personality Traits
提出CSI框架替代BFI，更可靠地评估大模型人格特质，为AI行为研究提供新标尺。
arXiv:2503.20182v2 Announce Type: replace-cross Abstract: As large language models (LLMs) increasingly function as human-like assistants exhibiting hu…
16
From Sequence to Structure: Relational Uncertainty Propagation for LLM Agents
不再只关注序列预测，而是引入关系不确定性传播，让LLM智能体在结构化决策中更可靠，值得深入研读。
arXiv:2608.16002v1 Announce Type: cross Abstract: Reliable uncertainty quantification (UQ) is essential for deploying large language model (LLM) agent…
17
Why HTTP 429 Should Not Count Against a Job's Retry Budget
别再让限流响应耗尽重试次数，把429当延迟而不是失败，任务系统可靠性的关键设计。
Most retry implementations—including BullMQ, custom wrappers, and managed queues—treat every non-2xx HTTP response as an identical failure. Each attem…
18
Your agent's "not done" lies as often as its "done"
别轻信AI助手的“已完成”，它可能和“未完成”一样爱说谎，亲手核对产物才是硬道理
There is a well-worn rule for running agents: don't trust the agent when it says it finished. Go look at the artifact. Good rule. We follow it. It did…
19
44.3 万块硬盘数据研究：HGST 可靠性最佳，东芝表现最差
IT之家 8 月 16 日消息，埃塞克斯大学经济学家克里斯托夫 · 西姆罗特（Christoph Siemroth）与成均馆大学经济学家朴汝明（Yeomyung Park）在 IEEE 发表了一项经同行评审的研究。该研究重新统计分析了云存储服务商 Backblaze 公开的 12 年硬盘运营数据，时…
20
Pet owners say smart pet feeder outage led to furry ones going unfed
智能喂食器一断网，毛孩子竟挨饿，智能硬件可靠性再遭拷问
"I've lost trust ..."