2026-07-23 科技日报

扫描 568 篇候选内容 · 覆盖 650 个信息源 · 纳入 556 篇

📌 今日要点

• Accelerating the frontiers of scientific discovery: Google’s $40M commitment to the Genesis Mission
• FALCON-Discover: Discovering Concentrated False-Confidence Regions for Calibration
• Beyond Output-Space Calibration: Spectral Evidence Bundling for Selective Reliability Estimation in Time-Series Classification
• Beyond Single-Dimensional Compression: The Compound Sparsity Frontier of Large Language Models
• ALAS: Additive Learnable Alpha-Stable Kernels for Flexible Bayesian Optimization
• FedCC: A Low-Resource Federated Adaptation of Foundation Models for Robust Corpus Callosum localization in Fetal Ultrasound Images

📝 科技简讯

• Accelerating the frontiers of scientific discovery: Google’s $40M commitment to the Genesis Mission — Google DeepMind
• FALCON-Discover: Discovering Concentrated False-Confidence Regions for Calibration — ArXiv ML (cs.LG)
• Beyond Output-Space Calibration: Spectral Evidence Bundling for Selective Reliability Estimation in Time-Series Classification — ArXiv ML (cs.LG)
• Beyond Single-Dimensional Compression: The Compound Sparsity Frontier of Large Language Models — ArXiv ML (cs.LG)
• ALAS: Additive Learnable Alpha-Stable Kernels for Flexible Bayesian Optimization — ArXiv ML (cs.LG)
• FedCC: A Low-Resource Federated Adaptation of Foundation Models for Robust Corpus Callosum localization in Fetal Ultrasound Images — ArXiv ML (cs.LG)
• Compressing What Matters: Neuron Importance Meets Data-Aware Low Rank Approximation for Language Model Compression — ArXiv ML (cs.LG)
• Edge-Efficient Transformer for End-to-End RF Spectrum Monitoring — ArXiv ML (cs.LG)
• 3 Google updates from Galaxy Unpacked 2026 — Google AI Blog
• A Multi-Agent Extension for Petri — LessWrong
• In other words: The influence of prompt variation on alignment evals — LessWrong
• Two Coefficient Givings beat one twice as big — LessWrong
• Can an LLM make a feature-length movie on its own? — LessWrong
• Will almost all future companies eventually be founded and run by autonomous AIs? — LessWrong
• AI Teammates: how monday.com runs production AI agents on Amazon Bedrock — AWS Machine Learning
• Kaggle + Google’s Free 5-Day Agentic AI Course — KDnuggets
• 10 Newsletters Keeping You Ahead in AI — KDnuggets
• 全球最高自由度！他们把人类身体「像素级」复刻了 — 新智元
• 刚刚，全球三大 AI 包揽 IMO 满分！击败 99% 人类 — 新智元
• 视频世界模型推理最高提速 2.59×！浙大新作无需重训，不改参数 — 新智元
• 中国黑马甩出 5 个模型、17 项全球第一！自进化体系杀进具身智能核心圈 — 新智元
• 纠偏率 98.15%！方向感觉醒，第一视角认路的国产 AI 开挂了 — 新智元
• [Paper] Stringological sequence prediction II — AI Alignment Forum
• Open models recap: more on Kimi K3, Qwen 3.8, Xi’s WAIC speech, distillation, the open-closed gap, and what’s next — Interconnects (Nathan Lambert)
• After shocking quarter, IBM insists that AI isn’t killing the mainframe — TechCrunch AI
• Build vs. Buy CRM in 2026: The True Cost Comparison — Salesforce AI Blog
• What Are the Best Healthcare CRMs for Growing Practices? — Salesforce AI Blog
• SysAdmin: Measuring Instrumental Power-Seeking in Frontier AI — arXiv AI (cs.AI)
• Calibrated Selective Fact-Checking via Evidence Chain Evaluation — arXiv AI (cs.AI)
• BatchDAG: LLM-Planned Execution Graphs for Scalable Ad-Hoc Analysis Over Enterprise Data — arXiv AI (cs.AI)
• AI Tool Discovery at Scale: All You Need is DNS — arXiv AI (cs.AI)
• From Agent Failure Paths to Quantified Residual Risk: A Compositional Framework for Resilient Agentic AI — arXiv AI (cs.AI)
• Anthropic’s $1.5B piracy settlement with book authors is a record loss that hands AI labs their biggest legal win — The Decoder
• Anthropic will deploy 2 gigawatts of AMD GPUs for Claude in a deal worth up to $5 billion — The Decoder
• Every frontier AI model tested by Britain’s safety institute tried to cheat on cybersecurity evaluations — The Decoder
• Cisco bets its small open cybersecurity models can outperform GPT-5.5 at vulnerability detection for a fraction of the cost — The Decoder
• OpenAI’s “Project Camellia” in Georgia secures a massive 3.2-gigawatt power deal through 2032 — The Decoder
• Samsung deepens its AI empire with a potential billion-euro stake in Europe’s hottest AI startup — The Decoder
• OpenAI claims responsibility for the Hugging Face hack after its own models escaped a test sandbox — The Decoder
• Building AI infrastructure with the Effingham County community — OpenAI Blog

📊 数据概览