https://www.sipply.dev/story/ai-startups-like-relay-and-character-ai-shut-down-amid-6c80c2d9 Sipply en 2026-10-24T19:00:00.000Z AI startups like Relay and Character.ai shut down amid industry challenges https://www.sipply.dev/zh/story/ai-startups-like-relay-and-character-ai-shut-down-amid-6c80c2d9 Sipply zh-cn 2026-10-24T19:00:00.000Z Relay 和 Character.ai 等人工智能初创公司在行业挑战中关闭 https://www.sipply.dev/story/pennsylvania-enacts-law-requiring-labels-on-ai-generated-db33acbd Sipply en 2026-10-24T18:32:21.000Z Pennsylvania enacts law requiring labels on AI-generated advertisements https://www.sipply.dev/zh/story/pennsylvania-enacts-law-requiring-labels-on-ai-generated-db33acbd Sipply zh-cn 2026-10-24T18:32:21.000Z 宾夕法尼亚州颁布法律，要求对AI生成的广告进行标记 https://www.sipply.dev/story/block-launches-dessa-a-human-like-voice-model-for-local-0500a4bc Sipply en 2026-10-24T16:00:00.000Z Block launches Dessa, a human-like voice model for local hardware https://www.sipply.dev/zh/story/block-launches-dessa-a-human-like-voice-model-for-local-0500a4bc Sipply zh-cn 2026-10-24T16:00:00.000Z Block推出Dessa，一个用于本地硬件的人性化语音模型 https://www.sipply.dev/story/groks-ai-model-favors-republicans-92-of-the-time-openais-gpt-f230ae28 Sipply en 2026-10-24T15:00:00.000Z Grok's AI model favors Republicans 92% of the time, OpenAI's GPT-6.1 favors Democrats 97% https://www.sipply.dev/zh/story/groks-ai-model-favors-republicans-92-of-the-time-openais-gpt-f230ae28 Sipply zh-cn 2026-10-24T15:00:00.000Z Grok的AI模型92%的时间偏向共和党，OpenAI的GPT-6.1偏向民主党97% https://www.sipply.dev/story/new-ai-agents-for-text-messaging-enhance-task-management-40c22ebd Sipply en 2026-10-24T14:00:00.000Z New AI agents for text messaging enhance task management capabilities https://www.sipply.dev/zh/story/new-ai-agents-for-text-messaging-enhance-task-management-40c22ebd Sipply zh-cn 2026-10-24T14:00:00.000Z 新的短信AI助手增强了任务管理能力 https://www.sipply.dev/story/nvidia-ceo-showcases-microduck-robots-leading-to-20-000-20136a13 Sipply en 2026-10-23T21:59:19.000Z Nvidia CEO showcases Microduck robots, leading to 20,000 preorders in one day https://www.sipply.dev/zh/story/nvidia-ceo-showcases-microduck-robots-leading-to-20-000-20136a13 Sipply zh-cn 2026-10-23T21:59:19.000Z Nvidia CEO展示Microduck机器人，单日预售达20,000台 https://www.sipply.dev/story/block-launches-dessa-an-ai-model-for-human-like-conversation-551a17a0 Sipply en 2026-10-23T19:32:39.000Z Block launches Dessa, an AI model for human-like conversation https://www.sipply.dev/zh/story/block-launches-dessa-an-ai-model-for-human-like-conversation-551a17a0 Sipply zh-cn 2026-10-23T19:32:39.000Z Block推出Dessa，一款用于人类对话的AI模型 https://www.sipply.dev/story/techcrunch-founder-summit-2026-to-feature-leaders-from-toast-8e1b0669 Sipply en 2026-10-23T19:06:23.000Z TechCrunch Founder Summit 2026 to feature leaders from Toast and Bessemer https://www.sipply.dev/zh/story/techcrunch-founder-summit-2026-to-feature-leaders-from-toast-8e1b0669 Sipply zh-cn 2026-10-23T19:06:23.000Z TechCrunch创始人峰会2026将邀请Toast和Bessemer的领导者 https://www.sipply.dev/story/the-iphone-duo-receives-a-7-10-review-sparking-debate-on-its-96322768 Sipply en 2026-10-23T17:00:18.000Z The iPhone Duo receives a 7/10 review, sparking debate on its identity as a product https://www.sipply.dev/zh/story/the-iphone-duo-receives-a-7-10-review-sparking-debate-on-its-96322768 Sipply zh-cn 2026-10-23T17:00:18.000Z iPhone Duo获得7/10的评价，引发关于其产品身份的争论 https://www.sipply.dev/story/openai-scales-habitat-storage-system-to-support-over-1-28907a3a Sipply en 2026-10-23T17:00:00.000Z OpenAI scales Habitat storage system to support over 1 billion ChatGPT users https://www.sipply.dev/zh/story/openai-scales-habitat-storage-system-to-support-over-1-28907a3a Sipply zh-cn 2026-10-23T17:00:00.000Z OpenAI扩展Habitat存储系统以支持超过10亿ChatGPT用户 https://www.sipply.dev/story/google-updates-privacy-policy-to-allow-gmail-and-photos-data-e0d96343 Sipply en 2026-10-23T16:52:55.000Z Google updates privacy policy to allow Gmail and Photos data for targeted ads https://www.sipply.dev/zh/story/google-updates-privacy-policy-to-allow-gmail-and-photos-data-e0d96343 Sipply zh-cn 2026-10-23T16:52:55.000Z 谷歌更新隐私政策，允许使用Gmail和Photos数据进行定向广告 https://www.sipply.dev/story/litpivot-launches-to-enhance-research-idea-development-d44b9690 Sipply en 2026-10-23T16:15:50.000Z LitPivot launches to enhance research idea development through literature contextualization https://www.sipply.dev/zh/story/litpivot-launches-to-enhance-research-idea-development-d44b9690 Sipply zh-cn 2026-10-23T16:15:50.000Z LitPivot推出，通过文献情境化增强研究想法的发展 https://www.sipply.dev/story/meta-postpones-new-business-terms-to-address-publisher-4e0856c4 Sipply en 2026-10-23T14:47:41.000Z Meta postpones new business terms to address publisher concerns over AI scraping https://www.sipply.dev/zh/story/meta-postpones-new-business-terms-to-address-publisher-4e0856c4 Sipply zh-cn 2026-10-23T14:47:41.000Z Meta推迟新商业条款以解决出版商对AI抓取的担忧 https://www.sipply.dev/story/ai-ports-retro-games-to-new-platforms-sparking-community-13a9bc5f Sipply en 2026-10-23T12:00:00.000Z AI ports retro games to new platforms, sparking community concerns https://www.sipply.dev/zh/story/ai-ports-retro-games-to-new-platforms-sparking-community-13a9bc5f Sipply zh-cn 2026-10-23T12:00:00.000Z 人工智能将复古游戏移植到新平台，引发社区担忧 https://www.sipply.dev/story/openai-enhances-security-for-frontier-model-training-with-fda34308 Sipply en 2026-10-23T10:00:00.000Z OpenAI secures frontier model training by moving to offline environments https://www.sipply.dev/zh/story/openai-enhances-security-for-frontier-model-training-with-fda34308 Sipply zh-cn 2026-10-23T10:00:00.000Z OpenAI通过转向离线环境来保障前沿模型训练 https://www.sipply.dev/story/study-finds-knowledge-distillation-excels-in-model-5a3613f1 Sipply en 2026-10-23T04:00:00.000Z Study finds knowledge distillation excels in model compression for edge platforms https://www.sipply.dev/zh/story/study-finds-knowledge-distillation-excels-in-model-5a3613f1 Sipply zh-cn 2026-10-23T04:00:00.000Z 研究发现知识蒸馏在边缘平台的模型压缩中表现优异 https://www.sipply.dev/story/researchers-introduce-moezip-for-improved-kv-cache-9066a401 Sipply en 2026-10-23T04:00:00.000Z Researchers introduce MoEZip for improved KV cache compression in LLMs https://www.sipply.dev/zh/story/researchers-introduce-moezip-for-improved-kv-cache-9066a401 Sipply zh-cn 2026-10-23T04:00:00.000Z 研究人员推出MoEZip，以改善LLM中的KV缓存压缩 https://www.sipply.dev/story/study-benchmarks-handshake-overhead-in-ai-agent-551636ec Sipply en 2026-10-23T04:00:00.000Z Study benchmarks handshake overhead in AI agent communication protocols https://www.sipply.dev/zh/story/study-benchmarks-handshake-overhead-in-ai-agent-551636ec Sipply zh-cn 2026-10-23T04:00:00.000Z 研究基准测试AI代理通信协议中的握手开销 https://www.sipply.dev/story/new-method-enhances-visual-mathematical-reasoning-in-vision-c28eb139 Sipply en 2026-10-23T04:00:00.000Z New method enhances visual mathematical reasoning in vision-language models https://www.sipply.dev/zh/story/new-method-enhances-visual-mathematical-reasoning-in-vision-c28eb139 Sipply zh-cn 2026-10-23T04:00:00.000Z 新方法增强了视觉语言模型中的视觉数学推理能力 https://www.sipply.dev/story/study-reveals-limitations-in-knowledge-tracing-models-for-2b782148 Sipply en 2026-10-23T04:00:00.000Z Study reveals limitations in knowledge tracing models for item parameter recovery https://www.sipply.dev/zh/story/study-reveals-limitations-in-knowledge-tracing-models-for-2b782148 Sipply zh-cn 2026-10-23T04:00:00.000Z 研究揭示知识追踪模型在项目参数恢复方面的局限性 https://www.sipply.dev/story/study-reveals-impact-of-prompt-style-on-gpt-5-5s-performance-8c94dd83 Sipply en 2026-10-23T04:00:00.000Z Study reveals impact of prompt style on GPT-5.5's performance in programming tasks https://www.sipply.dev/zh/story/study-reveals-impact-of-prompt-style-on-gpt-5-5s-performance-8c94dd83 Sipply zh-cn 2026-10-23T04:00:00.000Z 研究揭示提示风格对GPT-5.5在编程任务中的表现影响 https://www.sipply.dev/story/new-workflow-enhances-causal-identification-in-multimodal-bb8b0154 Sipply en 2026-10-23T04:00:00.000Z New workflow enhances causal identification in multimodal health records analysis https://www.sipply.dev/zh/story/new-workflow-enhances-causal-identification-in-multimodal-bb8b0154 Sipply zh-cn 2026-10-23T04:00:00.000Z 新工作流程增强了多模态健康记录分析中的因果识别 https://www.sipply.dev/story/new-method-enhances-neuro-symbolic-learning-with-soft-41f1bb81 Sipply en 2026-10-23T04:00:00.000Z New method enhances neuro-symbolic learning with soft ontology priors https://www.sipply.dev/zh/story/new-method-enhances-neuro-symbolic-learning-with-soft-41f1bb81 Sipply zh-cn 2026-10-23T04:00:00.000Z 新方法通过软本体先验增强神经符号学习 https://www.sipply.dev/story/study-recommends-hybrid-architecture-for-ai-agents-using-mcp-3ee5614b Sipply en 2026-10-23T04:00:00.000Z Study recommends hybrid architecture for AI agents using MCP and CLI strategies https://www.sipply.dev/zh/story/study-recommends-hybrid-architecture-for-ai-agents-using-mcp-3ee5614b Sipply zh-cn 2026-10-23T04:00:00.000Z 研究推荐使用MCP和CLI策略的混合架构用于AI代理 https://www.sipply.dev/story/new-spectral-framework-improves-uncertainty-quantification-8b6c19f9 Sipply en 2026-10-23T04:00:00.000Z New spectral framework improves uncertainty quantification in Bayesian learning https://www.sipply.dev/zh/story/new-spectral-framework-improves-uncertainty-quantification-8b6c19f9 Sipply zh-cn 2026-10-23T04:00:00.000Z 新的谱框架改善了贝叶斯学习中的不确定性量化 https://www.sipply.dev/story/researchers-unveil-roleflow-a-self-evolving-multi-role-5237c972 Sipply en 2026-10-23T04:00:00.000Z Researchers unveil RoleFlow, a self-evolving multi-role agentic system https://www.sipply.dev/zh/story/researchers-unveil-roleflow-a-self-evolving-multi-role-5237c972 Sipply zh-cn 2026-10-23T04:00:00.000Z 研究人员揭示了RoleFlow，一个自我进化的多角色代理系统 https://www.sipply.dev/story/new-ai-framework-speeds-up-wildfire-hazard-mapping-in-canada-89dc29cc Sipply en 2026-10-23T04:00:00.000Z New AI framework speeds up wildfire hazard mapping in Canada https://www.sipply.dev/zh/story/new-ai-framework-speeds-up-wildfire-hazard-mapping-in-canada-89dc29cc Sipply zh-cn 2026-10-23T04:00:00.000Z 新的AI框架加速加拿大的野火危险映射 https://www.sipply.dev/story/indrive-plans-2027-robotaxi-tests-in-kazakhstan-with-pony-ai-feca2e62 Sipply en 2026-10-23T04:00:00.000Z inDrive plans 2027 robotaxi tests in Kazakhstan with Pony.ai and Nvidia https://www.sipply.dev/zh/story/indrive-plans-2027-robotaxi-tests-in-kazakhstan-with-pony-ai-feca2e62 Sipply zh-cn 2026-10-23T04:00:00.000Z inDrive计划在2027年与Pony.ai和Nvidia在哈萨克斯坦进行机器人出租车测试 https://www.sipply.dev/story/ganet-achieves-99-99-accuracy-in-plant-disease-392d60b8 Sipply en 2026-10-23T04:00:00.000Z GANet achieves 99.99% accuracy in plant disease identification with minimal parameters https://www.sipply.dev/zh/story/ganet-achieves-99-99-accuracy-in-plant-disease-392d60b8 Sipply zh-cn 2026-10-23T04:00:00.000Z GANet在植物疾病识别中以最少的参数实现99.99%的准确率 https://www.sipply.dev/story/new-neural-network-architecture-enhances-uncertainty-dd1cbef1 Sipply en 2026-10-23T04:00:00.000Z New neural network architecture enhances uncertainty handling with belief-function layers https://www.sipply.dev/zh/story/new-neural-network-architecture-enhances-uncertainty-dd1cbef1 Sipply zh-cn 2026-10-23T04:00:00.000Z 新的神经网络架构通过信念函数层增强不确定性处理 https://www.sipply.dev/story/new-tool-assesses-privacy-risks-in-7-789-medical-ai-models-8ecc502b Sipply en 2026-10-23T04:00:00.000Z New tool assesses privacy risks in 7,789 medical AI models https://www.sipply.dev/zh/story/new-tool-assesses-privacy-risks-in-7-789-medical-ai-models-8ecc502b Sipply zh-cn 2026-10-23T04:00:00.000Z 新工具评估7,789个医疗AI模型的隐私风险 https://www.sipply.dev/story/researchers-develop-route-and-generate-framework-to-enhance-ebf00db7 Sipply en 2026-10-23T04:00:00.000Z Researchers develop Route-and-Generate framework to enhance LLM reasoning https://www.sipply.dev/zh/story/researchers-develop-route-and-generate-framework-to-enhance-ebf00db7 Sipply zh-cn 2026-10-23T04:00:00.000Z 研究人员开发路由与生成框架以增强LLM推理 https://www.sipply.dev/story/study-reveals-llm-agents-often-repeat-failed-moves-in-games-6253d4ba Sipply en 2026-10-23T04:00:00.000Z Study reveals LLM agents often repeat failed moves in games despite feedback https://www.sipply.dev/zh/story/study-reveals-llm-agents-often-repeat-failed-moves-in-games-6253d4ba Sipply zh-cn 2026-10-23T04:00:00.000Z 研究显示，LLM代理在游戏中经常重复失败的动作，尽管有反馈 https://www.sipply.dev/story/researchers-unveil-latent-interconnect-protocol-to-enhance-dba0d21f Sipply en 2026-10-23T04:00:00.000Z Researchers unveil Latent Interconnect Protocol to enhance large language models https://www.sipply.dev/zh/story/researchers-unveil-latent-interconnect-protocol-to-enhance-dba0d21f Sipply zh-cn 2026-10-23T04:00:00.000Z 研究人员揭示潜在互连协议，以增强大型语言模型 https://www.sipply.dev/story/new-human-ai-method-breaks-polynomial-construction-limits-in-efe5de45 Sipply en 2026-10-23T04:00:00.000Z New human-AI method breaks polynomial construction limits in Lehmer's problem https://www.sipply.dev/zh/story/new-human-ai-method-breaks-polynomial-construction-limits-in-efe5de45 Sipply zh-cn 2026-10-23T04:00:00.000Z 新的人类-AI方法打破Lehmer问题中的多项式构造限制 https://www.sipply.dev/story/mits-eko-ai-agents-invent-their-own-challenges-during-30-ff7be13b Sipply en 2026-10-23T00:45:37.000Z MIT's Eko AI agents invent their own challenges during 30-hour exploration https://www.sipply.dev/zh/story/mits-eko-ai-agents-invent-their-own-challenges-during-30-ff7be13b Sipply zh-cn 2026-10-23T00:45:37.000Z 麻省理工学院的Eko AI代理在30小时探索中发明自己的挑战 https://www.sipply.dev/story/hugging-face-launches-tool-to-check-model-compatibility-with-14c9ee10 Sipply en 2026-10-23T00:00:00.000Z Hugging Face launches tool to check model compatibility with AWS Trainium https://www.sipply.dev/zh/story/hugging-face-launches-tool-to-check-model-compatibility-with-14c9ee10 Sipply zh-cn 2026-10-23T00:00:00.000Z Hugging Face 推出工具以检查模型与 AWS Trainium 的兼容性 https://www.sipply.dev/story/openai-partners-with-polands-nask-to-enhance-ai-safety-and-69fac057 Sipply en 2026-10-23T00:00:00.000Z OpenAI partners with Poland's NASK to enhance AI safety and security https://www.sipply.dev/zh/story/openai-partners-with-polands-nask-to-enhance-ai-safety-and-69fac057 Sipply zh-cn 2026-10-23T00:00:00.000Z OpenAI与波兰NASK合作提升AI安全与保障 https://www.sipply.dev/story/thinkingbox-boosts-qwen3-8-27b-model-performance-with-rl-1efafe78 Sipply en 2026-10-22T22:24:34.000Z ThinkingBox boosts Qwen3.8-27B model performance with RL fine-tuning https://www.sipply.dev/zh/story/thinkingbox-boosts-qwen3-8-27b-model-performance-with-rl-1efafe78 Sipply zh-cn 2026-10-22T22:24:34.000Z ThinkingBox通过强化学习微调提升Qwen3.8-27B模型性能 https://www.sipply.dev/story/openai-aims-to-raise-30-billion-at-a-1-25-trillion-valuation-3c6d0f44 Sipply en 2026-10-22T22:08:24.000Z OpenAI aims to raise $30 billion at a $1.25 trillion valuation ahead of IPO https://www.sipply.dev/zh/story/openai-aims-to-raise-30-billion-at-a-1-25-trillion-valuation-3c6d0f44 Sipply zh-cn 2026-10-22T22:08:24.000Z OpenAI旨在以1.25万亿美元估值筹集300亿美元，准备IPO