Every story we have kept, newest first.
Looking for the daily editions? → Past editions
Page 7 of 1372
OpenAI announced that an unreleased internal model had proved the occurrence of singularities in the Navier–Stokes equations, one of the Millennium Prize Problems. A mathematician who had shared prior research alleges that OpenAI adopted a highly similar method, prompting a dispute over attribution and the validity of the result.
ITmedia AI+ (日本) · 8d ago
Kioxia’s top executive played down the prospect of deeper cooperation with stakeholder SK Hynix. The company also pledged to limit rising memory prices to protect long-term demand for AI infrastructure.
Bloomberg Technology · 8d ago
OpenAI launched ChatGPT Images 2.5, its most advanced image-generation model, with latency reduced by up to 50%. The update adds Sketch references, creative templates, image annotations, and two API models for different generation and editing workflows.
极客公园 · 8d ago
The article argues that AI adoption is moving from easy, creative, online tasks into harder traditional-enterprise workflows. Salesforce, Anthropic, OpenAI, Tencent, ByteDance, and Alibaba are positioning platforms, enterprise software, and field engineering as critical infrastructure for the next phase of AI deployment.
虎嗅 · 8d ago
HFresh is Weaviate's disk-based vector index for memory-efficient vector search. It combines low heap usage with incremental background maintenance.
Weaviate Blog · 8d ago
iOS 27 will bring changes to Apple HomeKit and the Home app as part of Apple’s broader AI revamps. The article focuses on the major improvements coming to Apple’s smart-home platform.
Engadget · 8d ago
The article argues that strong AI systems must learn from reality without optimizing for every signal it produces. Effective feedback loops require selective learning and the deliberate rejection of harmful or misleading objectives.
钛媒体 · 8d ago
Runway's annual recurring revenue has surpassed $200 million, with more than $40 million added in the second quarter alone. The company's strong revenue per employee highlights the momentum of AI video while its strategy shifts toward world models.
钛媒体 · 8d ago
ASML reportedly secured commitments from TSMC and Samsung to use its next-generation EUV lithography machines. The broader roundup also highlights rising Chinese auto exports and strategic NAND supply agreements involving Apple and Kioxia.
钛媒体 · 8d ago
Google reportedly obtained a $1.9 billion loan to restart a nuclear power plant for AI electricity demand, while its TPUv7 is said to deliver 50% higher inference performance than Nvidia. The briefing also covers AlphaGenome Atlas, Microsoft Project Opal, OpenAI’s Malaysian compute factory, and shifting chatbot market share.
钛媒体 · 8d ago
The Docker Blog explains six benefits of sandbox environments, including isolation, definable controls, and secure secrets and credential handling. Docker Sandboxes are presented as a way to create safer, more controlled environments for development and experimentation.
Docker Blog · 8d ago
The Linux Foundation opened a public comment period for SAFE, a guideline for sharing security incidents involving agentic AI across the industry. NVIDIA, Cisco, and CrowdStrike participated in developing the initial proposal.
ITmedia AI+ (日本) · 8d ago
Arm announced three new products covering mobile devices, AI infrastructure, and physical AI development. The launch includes Arm CSS for Mobile 2, Arm Neoverse CSS N4, and an expanded Arm Total Design service for physical AI.
ITmedia AI+ (日本) · 8d ago
The article examines claims that artificial intelligence helped solve the longstanding Navier–Stokes mathematics problem. It focuses on uncertainty surrounding the accomplishment and how credit should be attributed.
Engadget · 8d ago
SAP 正將認證考試從傳統知識測驗轉型為更重視實務能力的考核方式。部分新制考試允許使用 AI，考生需要展現實際解決問題與運用工具的能力。
ITmedia AI+ (日本) · 8d ago
越來越多求職者使用 ChatGPT 等生成式 AI 來諮詢企業選擇，AI 的回答可能影響投遞意願與面試後的選擇。企業因此需要關注自身在 AI 回應中的形象，並調整招聘與公開資訊策略。
ITmedia AI+ (日本) · 8d ago
Cisco President and Chief Product Officer Jeetu Patel discusses an open warning from Cisco, OpenAI, Anthropic, and others about increasingly sophisticated AI-enabled cyberattacks. He says AI tools can empower defenders and malicious actors alike, allowing attacks to operate at machine scale.
Bloomberg Technology · 8d ago
Indian workers are using iPhones to collect data that helps train humanoid robots. The report examines how the race for real-world training data could reshape labor markets and the future of work.
Bloomberg Technology · 8d ago
OpenAI claims that 10,000 of its AI systems solved the Navier–Stokes problem in 88 hours. The problem is one of the seven Millennium Prize Problems and has challenged mathematicians for nearly a century.
The Guardian Technology · 8d ago
Microsoft’s latest monthly patch release is unusually significant as security teams prepare for a potential surge in AI-assisted attacks. The article highlights the urgency of deploying patches before attackers can exploit vulnerable systems.
Ars Technica AI · 8d ago
A Claude user discovered that their account was consuming tokens while inactive. Anthropic has since warned users that hackers are targeting subscriber accounts.
TechCrunch AI · 8d ago
Cognition has reached a $48 billion valuation, suggesting investors see substantial room for multiple AI coding companies to compete. Its valuation multiple is reportedly higher than Cursor's before Cursor was sold to SpaceX.
TechCrunch AI · 8d ago
OpenAI claims its internal AI model discovered a solution to the long-unsolved Navier–Stokes problem, one of the seven Millennium Prize Problems. The system reportedly used a model more powerful than GPT-6 Astra and 10,000 concurrent agents, though the claim has generated controversy and requires independent verification.
The Verge · 8d ago
OpenAI released ChatGPT Images 2.5 and two related API models, with one already integrated into Adobe Firefly. The company also introduced prompt-sharing features as image creation across ChatGPT and its image API surpasses 3 billion images weekly.
The Next Web (TNW) · 8d ago
The US has accused Chinese AI firms of maliciously copying AI technology. Officials allege that these companies are conducting model-distillation activities at an industrial scale.
iTNews Australia · 8d ago
The federal government is proposing rules aimed at social media algorithms. The proposal would introduce greater user choice over how algorithmic systems operate or present content.
iTNews Australia · 8d ago
OpenAI has launched ChatGPT Images 2.5 with a new Sketch feature. Users can draw rough sketches inside ChatGPT and describe how they want the final image refined.
cnBeta (Full RSS) · 8d ago
OpenAI says an internal model proved that 3D Navier–Stokes equations can develop a finite-time singularity, using roughly 10,000 agents over 88 hours. The claim has not yet been accompanied by a public proof, while mathematicians Tristan Buckmaster and Levent Alpoge have published independently checkable Lean formalisations on related problems.
The Next Web (TNW) · 8d ago
Pathway describes how it develops and scales Baby Dragon Hatchling (BDH), a post-transformer architecture that reasons in latent space rather than generating chain-of-thought tokens. Its BDH-CQ variant reportedly achieved a new cost-efficiency mark on the ARC-AGI-1 benchmark.
AWS Machine Learning Blog · 8d ago
An unreleased OpenAI internal AI model reportedly solved the Navier–Stokes equations problem, with the result machine-verified. The claim remains unrecognized by the Clay Mathematics Institute as an official solution to a Millennium Prize Problem.
cnBeta (Full RSS) · 8d ago