Reflection与Nebius签下10亿美元算力大单，Nvidia追加20亿投资，AI基础设施赛道再掀巨浪。
Reflection AI has signed a $1 billion deal to access Nebius' compute. Reflection was founded in 2024 and is developing open source AI technology.
让你的AI助手通过屏幕快照读懂你的日常活动，开源项目Activity-frames把枯燥的数据转化为智能可理解的结构化帧。
Article URL: https://github.com/nossa-y/activity-frames Comments URL: https://news.ycombinator.com/item?id=48824758 Points: 1 # Comments: 0
AI生成同人文引发社区信任危机，粉丝用隐藏代码抓“作弊者”，一场关于创作伦理的战争正在上演。
Over the past week, a new fanworks movement has kicked off, with the aim to root out authors using generative AI. But the detection methods being impl…
《SOMA》精神续作《Ontos》跳票至2027，月球旅馆科幻悬疑延期登场
IT之家 7 月 4 日消息，在 TGA 2025 期间，曾打造《失忆症》与《活体脑细胞 SOMA》的恐怖游戏开发商 Frictional Games 揭晓了其最新力作《Ontos》，当时定于 2026 年发布，支持 PC、PlayStation 5、XBOX Series X|S 平台。 北京时间…
从网络平台责任豁免的历史教训，切入AI时代谁该为智能系统的行为负责这一关键议题。
Earlier this month, a German court ruled that Google is liable for its AI search summaries. Rejecting defenses like “users can check for themsel…
探讨AI生成虚构作品在真实世界中的形态与文化影响，一场AI与文学的跨界对话。
arXiv:2606.22748v1 Announce Type: cross Abstract: Some professional authors are beginning to use AI tools to help produce their fiction writing. Are r…
SpaceX与开源AI实验室Reflection签下每月1.5亿美元的超级计算合约，创下AI基础设施领域最大交易之一
Reflection AI will pay $150 million a month beginning July 1, 2026 through 2029 for immediate access to Nvidia's latest GB300 AI chips and supporting …
GitHub突然封禁开源组织CI，只因无关贡献者？企业级账号也难解禁。
A few weeks ago, Github decided to disable all Github Actions (including self-hosted runners) access for our open source org (lightningdevkit) for som…
Elastic豪掷8500万美元收购AI新星Deductive，背后是CRV和Databricks的押注。
DeductiveAI, a startup that uses AI to catch and resolve bugs in software, was founded just three years ago.
用几何感知的高斯面元实现主动式物体3D重建，兼顾效率与精度，为三维视觉研究提供新思路。
arXiv:2601.06997v2 Announce Type: replace-cross Abstract: Autonomous high-fidelity object reconstruction is fundamental for creating digital assets an…
新方法用鲁棒主动学习为Text-to-SQL精准挑选少样本示例，提升查询生成效率与泛化性。
arXiv:2606.10125v1 Announce Type: cross Abstract: Few-shot example retrieval is the dominant paradigm for grounding large language models (LLMs) in do…
突破线性限制，LLM行为控制新范式：可逆潜在变换实现更灵活激活引导
arXiv:2606.08454v1 Announce Type: new Abstract: Activation steering provides a lightweight inference-time mechanism for controlling large language mod…
多智能体协作的关键挑战：LLM如何分配子智能体所需知识？新基准PerspectiveGap来评测
arXiv:2606.08878v1 Announce Type: new Abstract: Real-world LLM applications are moving beyond single-agent workflows toward orchestrated multi-agent s…
面向异构网络威胁情报的RAG基准测试，填补安全领域大模型应用评估空白，KDD 2026收录。
arXiv:2510.11974v2 Announce Type: replace-cross Abstract: Cyber Threat Intelligence (CTI) is foundational to modern cybersecurity, enabling organizati…
新方法实现又快又忠实的函数向量，提升大模型行为控制的可信度与效率
arXiv:2606.05079v1 Announce Type: new Abstract: Function vectors (FVs) are task representations elicited during in-context learning that can be used t…
提出层级多智能体架构Multi²，基于LLM实现交互环境下的高效决策，为AI协作开辟新路径
arXiv:2606.03698v1 Announce Type: new Abstract: A central goal of large language model (LLM) research is to build agentic systems that can plan, act, …
用Interaction Tensor框架深挖知识蒸馏中学生模型的特征学习机制，揭示黑暗知识的秘密。
arXiv:2606.03052v1 Announce Type: new Abstract: Knowledge Distillation (KD) is a powerful tool for model compression, yet the precise mechanisms by wh…
用主动学习策略精准筛选高价值偏好数据，大幅降低RLHF数据标注成本，大模型偏好对齐的新效率方案。
arXiv:2603.09692v2 Announce Type: replace Abstract: Reinforcement Learning from Human Feedback (RLHF) has become the standard for aligning Large Langu…
用AI智能体自动生成基因组学交互式多视图可视化，突破传统手动创作瓶颈，提升科研效率。
arXiv:2606.00370v1 Announce Type: cross Abstract: Diverse genomics data, scientific questions, and analysis tasks typically demand highly specialized …
OpenAI联手日本云服务商Data Section，瞄准亚太企业级AI市场，加速全球化布局。
IT之家 5 月 30 日消息，日本云服务提供商 Data Section 昨天宣布与美国 AI 初创公司 OpenAI 达成战略合作，抢攻企业级 AI 市场。 据悉，Data Section 已在日本、泰国、马来西亚、澳大利亚和美国部署英伟达 GPU 集群数据中心，并将通过其企业级 AI 工作流平…