| 来源 | 时间 | 内容(英文原文 / 中文翻译) |
|---|---|---|
| X | 09-02 23:48 | Products I use that integrate AI perfectly (use regularly)
-Notion AI
-Spotify (AI DJ)
-Whoop
-Slack (Slackbot)
-X (Grok)
The ones that failed (never use):
-Instagram (Meta AI)
-Apple Intelligence
-DoorDash (AI chatbot)
-Gmail "Help me write
-Google Meet "take notes for me” 我日常使用、AI 集成得很好的产品：
-Notion AI
-Spotify（AI DJ）
-Whoop
-Slack（Slackbot）
-X（Grok）
失败的（几乎不用）：
-Instagram（Meta AI）
-Apple Intelligence
-DoorDash（AI 聊天机器人）
-Gmail「帮我写」
-Google Meet「帮我记笔记」
还有哪些？ @rowancheung · 查看原文 |
| X | 09-03 03:03 | Nobody can predict the future with perfect accuracy, but I am increasingly confident that by the end of decade, generative AI (at least we know it) will be viewed as on balance a tragic, costly mistake.
I hope this will not stop us from finding better ways to do AI — 没人能完美预测未来，但我越来越确信：到本十年末，生成式 AI（至少就我们现在所知的形态）会被总体视为一场悲剧性的、代价高昂的失误。
我希望这不会阻止我们在技术与政策上找到更好的 AI 路径，使其对社会产生更积极的净影响。 @GaryMarcus · 查看原文 |
| X | 09-03 01:31 | Had early access to Gemini 3.8 Flash, it is a very good Flash model, but not equivalent to a frontier model, though. Here is Gemini 3.8 Flash's version of the shader with the same prompts (which it provided very quickly) https://t.co/yXyH73BguR 提前体验了 Gemini 3.8 Flash，它是一款非常优秀的 Flash 模型，但还不能等同于前沿大模型。这是 Gemini 3.8 Flash 用相同提示词生成的着色器版本（生成速度很快）。 @emollick · 查看原文 |
| X | 09-03 01:53 | Grok @Bot now on Android! Grok @Bot 现已登陆 Android！ @elonmusk · 查看原文 |
| X | 09-03 00:44 | Introducing Gemini 3.8 Flash, another upgrade in under a month! And the new 3.8 Flash Cyber pushes the frontier of cyber defense. Relentless progress 🚀 More on the models & our cyber security work in the blog - happy building! https://t.co/C5h14qANIV 推出 Gemini 3.8 Flash，不到一个月又一次升级！全新的 3.8 Flash Cyber 将网络防御推向新前沿。持续推进。更多关于模型与网络安全工作的内容见博客——祝开发顺利！ @demishassabis · 查看原文 |
| X | 09-02 23:42 | Two new Gemini models are here to help scale your AI agents and secure code:
🔘 3.8 Flash: our most intelligent model yet with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.
🔘 3.8 Flash Cyber: our most capable https://t.co/EEJDIMhRwp 两款新的 Gemini 模型现已推出，助力扩展你的 AI 智能体并保障代码安全：
🔘 3.8 Flash：迄今最智能的 Flash 模型，相较 3.7 Flash 在软件工程、智能体任务与多步推理上均有显著提升。
🔘 3.8 Flash Cyber：能力最强的网络安全模型，具备前沿级漏洞检测与自动补丁能力。 @GoogleDeepMind · 查看原文 |
| X | 09-02 23:43 | Introducing Gemini 3.8 Flash, another jump in Gemini's agentic + coding capabilities, and our 3rd updated Flash model in only 6 weeks...
This model has been a ton of fun to work with, excited to see what you all think! https://t.co/Cj07lCBtp8 推出 Gemini 3.8 Flash，Gemini 在智能体与编程能力上又一次跃升，这也是我们仅 6 周内发布的第 3 个升级版 Flash 模型……
和这个模型一起工作非常有趣，期待大家的反馈！ @OfficialLoganK · 查看原文 |
| 来源 | 时间 | 内容(英文原文 / 中文翻译) |
|---|---|---|
| Hacker News 模型发布 | 09-02 23:12 | Gemini 3.8 Flash and 3.8 Flash Cyber Hacker News · 查看原文 |
| 新闻 应用落地 | 09-02 21:49 | Real-Time Intelligence with IBM Time Series Models on Confluent 基于 Confluent 上 IBM 时间序列模型的实时智能 |
| 新闻 安全伦理 | 09-02 23:40 | Proactive cyber defense for governments and enterprises 面向政府与企业的主动网络防御 |
| 新闻 模型发布 | 09-03 00:18 | Introducing Gemini 3.8 Flash and 3.8 Flash Cyber 隆重推出 Gemini 3.8 Flash 与 3.8 Flash Cyber |
| 新闻 安全伦理 | 09-03 00:24 | Proactive cyber defense for governments and enterprises 面向政府与企业的主动网络防御 |
| Hacker News 安全伦理 | 09-02 21:59 | Three sites made 215,128 “best software” pages for AI. Perplexity cites them Hacker News · 查看原文 |
| Hacker News 安全伦理 | 09-02 21:43 | Six curl CVEs after OpenAI and Anthropic came back with zero Hacker News · 查看原文 |
| Hacker News 其他 | 09-03 01:18 | Paint.net 5.2 alpha now runs on Linux Hacker News · 查看原文 |
| Hacker News 安全伦理 | 09-02 21:49 | A third of Perplexity's citations don't contain the number they're cited for Hacker News · 查看原文 |
| Hacker News 模型发布 | 09-03 03:49 | Fable 5.1 World Modeling Hacker News · 查看原文 |
| Hacker News 监管政策 | 09-03 04:57 | Mamdani Bans AI in NYC Schools Hacker News · 查看原文 |
| news.crunchbase. 资本动向 疑似·未核实 | 09-02 09:00 | 机器人感知公司 Lyte 融资 1.65 亿美元，投后估值达到 16 亿美元 这轮 C 轮融资由 Maverick Silicon 领投，使 Lyte 累计融资达到 2.72 亿美元；公司由前 Apple 和 PrimeSense 工程师创立，开发机器人使用的定制芯片、多模态传感器和空间感知软件。Crunchbase News、SiliconANGLE 及公司新闻稿相互印证。 news.crunchbase.com · 查看原文 |
| techcrunch.com 资本动向 疑似·未核实 | 09-02 08:01 | AI 安全公司 HiddenLayer 完成 1 亿美元 B 轮融资 本轮由 Delta-v Capital 领投，Ten Eleven Ventures、Morgan Stanley、微软 M12 和 Booz Allen Ventures 等参投；资金将用于模型、智能体及自主编码工具的运行时安全产品。TechCrunch 报道与 HiddenLayer 官方公告一致。 techcrunch.com · 查看原文 |
| blog.google 模型发布 疑似·未核实 | 09-02 | Google 发布 Gemini 3.8 Flash 与面向网络安全的 Flash Cyber 版本 Gemini 3.8 Flash 主攻软件工程、智能体和多步推理，首发价为每百万输入词元 0.75 美元、输出词元 3.75 美元；Flash Cyber 针对漏洞发现和缓解进行优化。Google 官方信息获 Ars Technica 与 9to5Google 独立验证。 blog.google · 查看原文 |
| research.meta.ai 模型发布 疑似·未核实 | 09-02 | Meta 发布 Muse Spark 1.3，重点提升长周期智能体与编码任务表现 Muse Spark 1.3 已上线 Muse Code 和 Meta Model API；Meta 称其在编码任务中较 1.2 减少约 20% 工具调用和约 25% 词元消耗，并增强提示注入防御与高后果操作确认机制。官方发布与 Axios 独立报道相互印证。 research.meta.ai · 查看原文 |
| Hacker News 安全伦理 | 09-03 00:58 | Mushroom hunting with LLMs: what can go wrong? Hacker News · 查看原文 |
| Hacker News 其他 | 09-02 21:03 | LLMs: Intelligence vs. Cost Hacker News · 查看原文 |
| Hacker News 其他 | 09-03 00:14 | Show HN: FrontierHarness Eval – 9 harness, same model, cost per pass varies 17x Hacker News · 查看原文 |
| Hacker News 其他 | 09-03 03:41 | The Post-AI Internet Doesn't Look Great Hacker News · 查看原文 |
| Hacker News 模型发布 | 09-02 23:44 | Gemini 3.8 Flash and 3.8 Flash Cyber Hacker News · 查看原文 |
| Hacker News 其他 | 09-03 02:42 | Qantas Airbus A380 engine failure in 2010 (2023) Hacker News · 查看原文 |
| Hacker News 安全伦理 | 09-03 07:08 | METR Report on OpenAI / Hugging Face Hacking Incident Hacker News · 查看原文 |
| Hacker News 应用落地 | 09-02 22:02 | WebLLM: high-performance in-browser LLM inference engine Hacker News · 查看原文 |
| 新闻 其他 | 09-02 20:07 | Blending Concepts: Benchmarking Visual Metaphor Generation in Text-to-Image Models Text-to-image (T2I) models have achieved remarkable success at faithfully rendering specified objects and attributes, yet their ability to produce visual metaphors, images that convey abstract ideas by combining elements from two distinct domains, remains largely unexamined. To bridge this gap, we introduce VMetaphor-Bench, the first benchmark for evaluating visual metaphor generation in T2I model |
| 新闻 其他 | 09-02 20:12 | Rethinking the Teacher-Student Framework for Test-Time Adaptation Test-Time Adaptation (TTA) has recently emerged as a promising strategy that allows the adaptation of pre-trained models to changing data distributions at deployment time, without access to any labels. To mitigate error accumulation, researchers have widely adopted the teacher-student framework, though its long-term stability is often taken for granted. In this work, we challenge the common strate |
| 新闻 应用落地 | 09-02 20:15 | Orthogonal Ensembles and Tested Explanations for Performer-Independent Body-Motion Emotion Recognition We study body-only, 12-class acted-emotion classification from skeleton motion under leave-performer-out (LPO) evaluation, a hard, underdetermined setting: chance is 8.3%, and a protocol-matched reproduced STGCN++ baseline reaches only 25.73 +/- 4.03% Macro-F1. We show that reliable gains come not from a new architecture but from combining eleven models with orthogonal error modes: under 10-fold L |
| 新闻 其他 | 09-02 20:25 | Spectral Initialization and Scheduled Graph Smoothness for Uncertain Knowledge Graph Completion Uncertain knowledge graphs (UKGs) extend knowledge graphs by assigning each triple a continuous confidence score. Since most possible triples lack observed confidences, recent methods rely on semi-supervised learning to generate pseudo-labels. These methods initialize entity embeddings without using the confidence-weighted graph, discarding its global community and hub structure. We introduce QUES |
| 新闻 其他 | 09-02 20:30 | When Persona Attributes Improve Population Alignment in Large Language Models Large Language Models (LLMs) are increasingly used to predict the responses of human participants in survey panels. Towards that goal, persona prompting has recently emerged as a technique to inform and align large pretrained language models. Persona prompting refers to the practice of using short textual descriptions of 'personas' in prompts to steer the LLM's generations. Personas describe indiv |
| 新闻 其他 | 09-02 20:37 | Fine-Grained Anomaly Perception in Wild UGC-Enhanced Images: A Comprehensive Dataset and Difference-Fusion Framework Image enhancement and restoration have become standard back-end operations on short-video and social media platforms to boost UGC visual experience. Yet these processes inevitably introduce visual anomalies--especially in faces, texts, and textures--that directly undermine perceptual fidelity and viewer trust. While existing IQA methods perform well on classic distortions, they target holistic qua |
| 新闻 应用落地 | 09-02 20:50 | A Comparative Study of Graph Representations for GNN-Based Power Grid Control in L2RPN Graph construction is a critical but underexamined design choice in deep reinforcement learning for power grid control. We present a controlled experimental comparison of different graph representations, including physical topology, electrical-sensitivity, and hybrid variants for topology control in the Learning to Run a Power Network (L2RPN) environment. Our findings indicate that matching graph |
| 新闻 应用落地 | 09-02 20:51 | TrajMind: Chaining Role-Specialized LoRAs for Fast-and-Slow Collective Trajectory Anomaly Diagnosis Diagnosing collective anomalies from urban trajectories is increasingly important for traffic governance, as it reveals what happened, who was involved, and where and when the event occurred. Existing detectors efficiently produce scores or labels, whereas vision--language pipelines provide richer semantics; neither couples verifiable diagnosis with low-latency monitoring. The central challenge is |
| 新闻 其他 | 09-02 21:00 | Learn from Whoever Is Right: Answer-Verified Multi-Teacher Distillation for Multi-Domain LLMs Modern large language models (LLMs) rely on reinforcement learning to build strong capabilities in individual domains, but integrating those capabilities into a single deployable model remains challenging. By routing each sample to the teacher whose domain matches it, existing approaches let a domain label decide which teacher provides supervision. However, domain expertise holds only on average: |
| 新闻 应用落地 | 09-02 21:00 | ProbeMatchDTI: Probe-Driven Multi-Scale Biochemical Pattern Matching for Drug-Target Interaction Prediction Drug-target interaction (DTI) prediction is an important task in AI-driven drug discovery. Although recent biochemical representation learning methods have improved DTI prediction, their passive feature aggregation tends to favor dominant molecular patterns while suppressing weak yet binding-relevant signals, such as functional groups and residue-context patterns, limiting the modeling of multi-sc |
| 新闻 应用落地 | 09-02 21:17 | Online Reinforcement Learning in the Met Office Unified Model through Distributed Model-Agent Coupling Machine-learnt corrections can complement numerical weather prediction only if they adapt to the evolving model state while preserving dynamical consistency and numerical stability. To test this within a global forecasting model, we couple the Met Office (UKMO) Unified Model (UM) with distributed RL agents through rank-local tensors. A DDPG actor shares weights across the 70 vertical model levels |
| 新闻 安全伦理 | 09-02 21:17 | Learning-Based Reconstruction Attacks on Coordinate-Obfuscated Point Clouds Volumetric video based on point cloud representations enables immersive virtual and augmented reality applications but introduces significant challenges for efficient and secure content delivery. Prior work proposed a selective coordinate encryption framework for point clouds that encrypts only a subset of coordinates, reducing computational costs while visually degrading unauthorized content. How |
| 新闻 其他 | 09-02 21:24 | Competitive Market Behavior of LLMs Large language models (LLMs) are increasingly deployed as economic agents, yet there is little evidence whether LLM agents are suited for participating in market mechanisms designed for humans, and whether these mechanisms deliver desired outcomes when faced with LLM agents. We address this question by replicating seminal economic experiments, replacing human subjects with LLM agents. We place age |
| 新闻 应用落地 | 09-02 21:48 | Predictors of Loneliness in Older Adults Using Multimodal Analysis of Speech and Language Loneliness is a critical public health issue among older adults, linked to higher risks of depression, cognitive decline, and mortality. Scalable, objective methods for its detection remain limited, particularly in natural conversational contexts. We analyzed speech and language markers of loneliness in 310 older adults using semi-structured telephone interviews to help understand how they process |
| 新闻 其他 | 09-02 21:59 | Collective creativity in hybrid societies Generative AI is changing how cultural artifacts are created and circulated, and with it our understanding of creativity itself. Researchers disagree about whether these tools enrich or impoverish culture, and we argue that much of that disagreement comes from conflating two distinct components of creativity: novelty, a property of single artifacts, and diversity, a property of populations. We arg |
| 新闻 其他 | 09-02 22:01 | Source Distribution Estimation by Posterior Averaging Simulation-based science often requires a distribution over simulator parameters whose push-forward reproduces a set of real observations: this is the source distribution estimation (SDE) problem. Existing methods fit the source against a likelihood surrogate trained once from a fixed proposal prior. Their objective is therefore stated only in terms of the surrogate instead of the true simulator, |
| 新闻 其他 | 09-02 22:02 | Scalable Direction-Following TTS via Voice Impression-Guided Pseudo Triplet Construction Voice actors often re-read the same script while modifying their delivery in response to performance directions. We study this setting as direction-following TTS, where a system generates a new utterance that reflects a given direction relative to a reference utterance while preserving speaker identity and linguistic content. A key challenge is the lack of training data capturing such relative mod |
| 新闻 安全伦理 | 09-02 22:03 | Automated Vulnerability Injection in Smart Contracts Using Large Language Models Assessing vulnerability detection tools for smart contracts requires datasets with known ground truth, yet such datasets are scarce and difficult to build by hand. We propose an approach that uses Large Language Models (LLMs) to automatically inject vulnerabilities into Solidity smart contracts, and demonstrate it in a case study targeting 49 vulnerability types from OpenSCV. Injected contracts ar |
| techcrunch.com?p 算力芯片 | Wed, 02 Sep 2026 | Jio’s pitch of an “AI-ready” PC through JioPC also challenges the conventional idea of what an AI PC looks like. The company has not disclosed much about the underlying hardware powering the cloud service. However, the specifications it has provided do not mention dedicated AI accelerators for on-device generative AI workloads, which is a key feature of the new generation of AI PCs. techcrunch.com?p=3160089 · 查看原文 |
| techcrunch.com 安全伦理 | Wed, 02 Sep 2026 | Under normal circumstances, a reasoning model’s chain of thought provides the sequential steps taken by the model as it attempts to solve a problem. While the representation is imperfect, it still serves as a valuable tool for monitoring misbehavior or misalignment. In the case of OpenAI’s recent rogue agent activity, chain-of-thought records were an important tool in teasing out why agents behave techcrunch.com · 查看原文 |
| techcrunch.com 资本动向 | Wed, 02 Sep 2026 | Founded in early 2025, Wonderful has grown blazingly fast. It initially saw strong traction with its customer service AI agent platform, which it built and customized to sell in various non-English-speaking markets. Its strategy of deploying engineering teams to work with its customers, sometimes on their premises, to deploy and integrate its AI tech into their workflows and systems also proved ve techcrunch.com · 查看原文 |
| openai.com 其他 | Wed, 02 Sep 2026 | Careers
## Product Engineer, Ona
Codex - Engineering - San Francisco and London, UK
About the Team
The Ona team at OpenAI is helping build the software factory for the enterprise. We build infrastructure that enables AI agents to work in secure, customer-controlled cloud environments, with the context, tools, and controls they need to make progress across the software lifecycle—beyond a single openai.com · 查看原文 |
| deepmind.google 模型发布 | Wed, 02 Sep 2026 | Sep 02, 2026
Our newest Gemini models deliver next-generation intelligence for agentic workflows and cybersecurity.
Tulsee Doshi
Senior Director, Product Management
Raluca Ada Popa
Gemini Security Lead, Google DeepMind
a hero image reading "Gemini 3.8 Flash and 3.8 Flash Cyber" [...] ## Related stories
### Proactive cyber defense for governments and enterprises
### The latest AI news we an deepmind.google · 查看原文 |
| techcrunch.com 资本动向 | Wed, 02 Sep 2026 | How quickly things have changed. Security companies are now scrambling to build products that can monitor not just agents but also the tools and add-ons they use. While there still aren’t many headlines about agents being exploited, the risk of agents going haywire during production is nevertheless real. And so the market for tools to prevent that from happening has exploded: Gartner estimates com techcrunch.com · 查看原文 |
| openai.com 其他 | Wed, 02 Sep 2026 | Success in this role means accelerating customer confidence, increasing product adoption, supporting successful launches of new product capabilities, and helping customers translate product features into tangible outcomes across teams and business functions. You will own outcomes related to activation and sustained usage by shaping how enablement drives measurable customer impact. [...] This role openai.com · 查看原文 |
| cdn.openai.com 其他 | Wed, 02 Sep 2026 | The same capabilities could change what it takes to start and run a business. Software development, fi nancial planning, and market analysis often require employees or outside providers. Agents could lower some of those costs. Two effects could follow:
● Existing fi rms could develop new products and enter new markets more quickly. Agents could help them test ideas, serve new customers, and ex cdn.openai.com · 查看原文 |
| theverge.com 其他 | Wed, 02 Sep 2026 | # Acer’s new MacBook Air rival is nearly a pound lighter than Apple’s laptop. The half-inch thin laptop is joined by a new, much larger Swift Air 16 — both with Intel Wildcat Lake. Acer is announcing a new laptop poised to rival the MacBook Air’s thinness and lightness — and in the process, it cops another laptop’s name. That’s around 1.7mm thicker than Apple’s svelte 13-inch MacBook Air, but Ace theverge.com · 查看原文 |
| techcrunch.com 应用落地 | Wed, 02 Sep 2026 | # PSA: Amazon’s shopping AI can now tell you if that message is a scam. The retailer said some 360,000 customers every year reach out to its customer service department wondering whether a message they received from Amazon is real or a scam. Now, customers will be able to ask Amazon’s consumer AI service, Alexa for Shopping, to help answer that question instead. The company said its AI can “defini techcrunch.com · 查看原文 |
| techcrunch.com 算力芯片 | Wed, 02 Sep 2026 | # India’s richest man now wants to turn aging computers into AI-ready PCs. Mukesh Ambani’s Reliance Jio is opening its cloud PC service to any internet user in India, positioning it as a way to turn aging computers into AI-ready machines without hardware upgrades. The service streams a virtual PC from Jio’s cloud to an existing computer, with plans offering up to eight virtual CPUs, 16GB of RAM, a techcrunch.com · 查看原文 |
| techcrunch.com 其他 | Wed, 02 Sep 2026 | # TechCrunch Disrupt 2026’s new Real World AI Stage features Nvidia, robots, and extinct animals. But the Real World AI Stage is brand-new for TechCrunch Disrupt 2026. On this stage, we’ll be focusing on that intersection between the digital and physical, and all the ways we’ll continue to see a blending of the two, as autonomous hardware goes beyond self-driving cars and enters public spaces, bat techcrunch.com · 查看原文 |
| techcrunch.com 资本动向 | Wed, 02 Sep 2026 | Adobe has acquired India-based marketing intelligence startup Rilo in a deal involving licensing and team acquisition, TechCrunch learned and the company confirmed. This is Adobe’s second acquisition from India after it bought video platform Rephrase.ai in 2023. The acquisition gives Adobe a small team and technology focused on automating marketing workflows, which have been changing as companies techcrunch.com · 查看原文 |
| techcrunch.com 资本动向 | Wed, 02 Sep 2026 | When AI security startup HiddenLayer raised its $50 million Series A three years ago, one of the big questions in the space was whether the AI threats the startup was protecting against would manifest in enough quantity to make for a real market. To make the most of that momentum, the startup has now raised $100 million in a Series B funding round that was led by Delta-v Capital, with participatio techcrunch.com · 查看原文 |
| techcrunch.com 资本动向 | Wed, 02 Sep 2026 | # Palo Alto Networks paid $500M for Thrive-backed Console, sources say. Palo Alto Networks paid $500 million in cash and stock to acquire Console, a two-year-old startup that uses AI agents to automate routine IT help desk tasks, according to two people with knowledge of the deal. Before the sale, Console was valued at $157 million, according to PitchBook, delivering a rapid return for investors i techcrunch.com · 查看原文 |
| techcrunch.com 资本动向 | Wed, 02 Sep 2026 | Israeli-Dutch AI startup Wonderful has raised $550 million in a Series C funding round that gives it a price tag of $5 billion, more than doubling the $2 billion valuation the company gained when it last raised funding nearly six months ago. Salesforce is investing in the company for the first time in this round. Departing from its customer service roots, Wonderful now offers a platform, dubbed “W techcrunch.com · 查看原文 |
| techcrunch.com 监管政策 | Wed, 02 Sep 2026 | In a lawsuit that The New York Times filed against OpenAI, the Trump administration has contributed a 20-page brief in defense of the ChatGPT maker’s unlicensed use of copyrighted material to train its LLMs. “The United States has a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets the standard for the practice and procedure of AI use glob techcrunch.com · 查看原文 |
| techcrunch.com 监管政策 | Wed, 02 Sep 2026 | In a lawsuit that The New York Times filed against OpenAI, the Trump administration has contributed a 20-page brief in defense of the ChatGPT maker’s unlicensed use of copyrighted material to train its LLMs. “The United States has a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets the standard for the practice and procedure of AI use glob techcrunch.com · 查看原文 |
| techcrunch.com 安全伦理 | Wed, 02 Sep 2026 | Edelson PC, the law firm that filed seven lawsuits against OpenAI in April on behalf of victims and families connected to the Tumbler Ridge mass shooting, is filing 30 additional complaints this week. The complaints come as OpenAI continues to deal with the fallout of another high-profile safety incident after one of the company’s AI models escaped its sandbox during cybersecurity evaluations and techcrunch.com · 查看原文 |
| theverge.com 安全伦理 | Wed, 02 Sep 2026 | As details about the model trickle out, researchers are warning it “may be the single worst development for AI security/safety to date.”. Shortly after OpenAI said on Tuesday that it had delayed Astra’s release to work on safety issues, *The Information* reported that Astra shows far less of its “thinking” than other frontier AI models, sparking concern it could be dangerously hard to monitor. Thi theverge.com · 查看原文 |
| arstechnica.com 模型发布 | Wed, 02 Sep 2026 | # Google releases Gemini 3.8 Flash, its third Flash model in six weeks. Google’s Pro model updates are seemingly paused, but there’s yet another Gemini Flash today. Today, Google is announcing its third Flash model release in just six weeks, making it more likely that we’ll never see the promised Gemini 3.5 Pro. But no matter, says Google, because Gemini 3.8 Flash is its best reasoning and coding arstechnica.com · 查看原文 |
| techcrunch.com 安全伦理 | Wed, 02 Sep 2026 | OpenAI’s new Astra model will use a reasoning technique called “recurrent depth” that allows it to operate outside of the sequential thinking that characterizes most reasoning models, the The Information reported on Tuesday. This technique, also called “opaque recurrence,” will likely make the model’s chain of thought more difficult to monitor — and that has AI safety experts rattled. While Astra’ techcrunch.com · 查看原文 |
| Hacker News 应用落地 | 09-03 01:04 | ChatGPT ad targeting is garbage Hacker News · 查看原文 |
| Hacker News 安全伦理 | 09-03 02:30 | Firefox's AI Switch Is Off. Telemetry Isn't Hacker News · 查看原文 |
| Hacker News 其他 | 09-03 00:25 | The race to engineer new knobs for the human brain Hacker News · 查看原文 |
| Hacker News 其他 | 09-03 04:08 | Altair Basic Interpreter Source Code (1975) [pdf] Hacker News · 查看原文 |
| Hacker News 应用落地 | 09-03 03:51 | AI Agents and the Refactoring That Never Happens Hacker News · 查看原文 |
| Hacker News 监管政策 | 09-02 23:36 | AI Policy Hacker News · 查看原文 |
| MarkTechPost 模型发布 | 09-03 01:20 | Google DeepMind Releases Gemini 3.8 Flash and Gemini 3.8 Flash Cyber: One Core Model, Two Access Envelopes - MarkTechPost 谷歌 DeepMind 发布 Gemini 3.8 Flash 与 Gemini 3.8 Flash Cyber：一个核心模型，两种访问封装 - MarkTechPost MarkTechPost · 查看原文 |
| forkast.news 模型发布 | 09-03 02:39 | Google DeepMind Ships Gemini 3.8 Flash and Cyber: Six Weeks, Three Flash Models, One Compute Landlord Thesis - forkast.news 谷歌 DeepMind 发布 Gemini 3.8 Flash 与 Cyber：六周、三款 Flash 模型、一个算力地主论题 - forkast.news forkast.news · 查看原文 |
| blog.google 模型发布 | 09-02 23:47 | Introducing Gemini 3.8 Flash and 3.8 Flash Cyber - blog.google 隆重推出 Gemini 3.8 Flash 与 3.8 Flash Cyber - blog.google blog.google · 查看原文 |
| automate.org 应用落地 | 09-02 22:03 | Industry Insights: Google DeepMind's Robotics Lead on Gemini Physical AI Push - automate.org 行业洞察：谷歌 DeepMind 机器人负责人谈 Gemini 具身智能推进 - automate.org automate.org · 查看原文 |
| GlobeNewswire 模型发布 | 09-03 03:43 | Multiverse Computing launches Quasar 438B, the - GlobeNewswire Multiverse Computing 推出 Quasar 438B，该 - GlobeNewswire GlobeNewswire · 查看原文 |
| Forbes 安全伦理 | 09-03 01:08 | 5 Ways AI Models Make Women Look Less Suited For Leadership - Forbes 人工智能模型让女性显得更不适合领导岗位的五种方式 - Forbes Forbes · 查看原文 |
| Tech Xplore 安全伦理 | 09-03 04:40 | Researchers develop cost-efficient method for detecting hallucinations in large language models - Tech Xplore 研究人员开发出检测大语言模型幻觉的低成本方法 - Tech Xplore Tech Xplore · 查看原文 |
| ms.now 安全伦理 | 09-02 21:40 | Judge slams HHS for referencing fake studies generated by misuse of AI - ms.now 法官批评卫生与公众服务部引用因滥用人工智能而产生的虚假研究 - ms.now ms.now · 查看原文 |
| TechCrunch 监管政策 | 09-03 01:09 | US government sides with OpenAI on issue of training LLMs on copyrighted material - TechCrunch 美国政府在利用版权材料训练大语言模型问题上站在 OpenAI 一边 - TechCrunch TechCrunch · 查看原文 |
| Vatican News 安全伦理 | 09-02 21:55 | AI will provide ‘an explosion of creativity, but brakes are needed’ - Vatican News 人工智能将带来‘创造力大爆发，但需要踩刹车’ - Vatican News Vatican News · 查看原文 |
| Politico 监管政策 | 09-03 02:33 | Trump administration backs OpenAI in newspapers' copyright case - Politico 特朗普政府在报纸版权案中支持 OpenAI - Politico Politico · 查看原文 |
| NBC News 其他 | 09-03 02:29 | JD Vance talks end times, Antichrist and AI on faith podcast - NBC News JD·万斯在信仰播客中谈及末日、敌基督与人工智能 - NBC News NBC News · 查看原文 |
| NYC.gov 监管政策 | 09-03 07:50 | Mayor Mamdani and Chancellor Samuels Put Students First with Nation’s Broadest Generative AI Moratorium in Schools - NYC.gov 市长马姆达尼与学监塞缪尔斯将学生放在首位，推出全美范围最广的校园生成式人工智能禁令 - NYC.gov NYC.gov · 查看原文 |
| Nature 应用落地 | 09-03 00:30 | Designing physics experiments with artificial intelligence - Nature 用人工智能设计物理实验 - Nature Nature · 查看原文 |
| 来源 | 市场 | 榜位 | 时间(北京) | 标题(点击查看原文) |
|---|---|---|---|---|
| 百度热搜 | 综合 | 9 | 09-03 09:15 | 仿生机器人女性偏多 |
| 知乎 | 综合 | 16 | 09-03 09:15 | 怎么看美国《外交政策》（Foreign Policy）称「中国的成功正迫使美国重新思考 AI 战略」？ |
| 知乎 | 综合 | 19 | 09-03 09:15 | Gemini 3.8 Flash 发布，有什么值得关注的？ |
| 微博 | 综合 | 20 | 09-03 09:15 | 戚薇回应做AI授权是因为缺钱 |