Every story we have kept, newest first.
Looking for the daily editions? → Past editions
Page 1097 of 1375
Meta's AI agent instructed an engineer on an internal forum to implement a solution that exposed sensitive user and company data to employees for two hours. The incident stemmed from a query for engineering guidance.
The Guardian Technology · 193d ago
US tech giant Palantir is suing small Swiss magazine Republik after journalists investigated its 'European hub' in Altendorf. The probe suggests Palantir has failed in Switzerland despite 2021 announcements.
The Guardian Technology · 193d ago
Debate rages on replacing supporting actors with AI in Chinese entertainment, but platforms prioritize star traffic. AI excels in backgrounds, VFX, and crowd fills, not acting due to performance needs and talent pipelines.
虎嗅 · 193d ago
SpaceX team ordered heterojunction PV equipment from a leading Chinese firm, with shipment expected in the first week of May. The export requires regulatory approval.
36氪 · 193d ago
Alibaba merges DingTalk into new Token Hub with Wukong as CLI-based enterprise AI OS for workflows. Emphasizes natural token use over forced consumption, supports multi-models and hardware.
虎嗅 · 193d ago
JPMorgan to pilot reports contrasting self-reported hours with AI-estimated ones for junior bankers, based on digital footprints. Data includes video calls, keyboard activity, and schedules.
36氪 · 193d ago
Former Honor AI lab director leads team deploying 'Super Brain' AI to manage farms autonomously around the clock. This system enables non-stop operations, drastically cutting costs by 60%.
量子位 · 193d ago
The White House plans to submit a draft AI regulatory framework to US Congress this Friday. It serves as a conceptual outline for federal AI oversight, balancing innovation promotion with risk mitigation.
cnBeta (Full RSS) · 193d ago
US Southern District of New York prosecutors charged three associates of a US server maker with illegally shipping Nvidia AI chip-equipped servers to China. The operation involved billions in value without authorization.
cnBeta (Full RSS) · 193d ago
Nanyang Technological University team adapts disaster-rescue cyborg cockroaches for infrastructure inspection. These bio-hybrids enter narrow spaces to check aging pipelines and underground facilities.
cnBeta (Full RSS) · 193d ago
Xiaomi launched its new-generation SU7 smart EV with standard edition at 219,900 yuan (US$31,900), pro at 249,900 yuan, and max at 303,900 yuan. Prices are 10,000 yuan lower than pre-sale estimates.
SCMP Technology · 193d ago
LibTV has launched as the first professional video creation platform designed for both human creators and AI agents. It enables seamless, end-to-end video production workflows.
Pandaily · 193d ago
NVIDIA CEO Jensen Huang calls Anthropic the AI programming leader with Claude topping all models, including OpenAI's. He forecasts $1 trillion revenue by 2030 and advises bolder ambitions.
cnBeta (Full RSS) · 193d ago
AISTALAND has unveiled its first model, the next-generation smart wagon GT7. The vehicle blends AI-driven intelligence with high-end design.
Pandaily · 193d ago
ModelBest has launched EdgeClaw Box, a secure, out-of-the-box AI agent hardware system. It processes sensitive data locally to ensure privacy.
Pandaily · 193d ago
Alibaba has unveiled its 'Token Strategy' to drive a token-based AI economy. The company aims for $100 billion in annual cloud and AI revenue within five years.
Pandaily · 193d ago
OpenAI plans a desktop superapp integrating ChatGPT, Codex, and browser for All-in-One access amid Anthropic rivalry. Acquisitions of Astral (dev tools), Promptfoo (safety), Torch (retrieval) enable agentic AI that autonomously operates on user PCs.
虎嗅 · 193d ago
ADASPACE has achieved the world's first demonstration of space-based AI computing directly powering and controlling ground robots on Earth. The experiment uses natural language for commands, marking a milestone in space-to-Earth AI integration.
Pandaily · 193d ago
A massive survey of 81,000 people uncovers late-night confessions about AI's impact. Participants fear AI is eroding humanity's most precious elements.
钛媒体 · 193d ago
Xiaohongshu bans AI tools like OpenClaw that fully automate accounts, simulating human content and interactions to protect 'live human sense'. It parallels pre-AI 'Zhengzhou gang' factories, now AI-enhanced.
虎嗅 · 193d ago
Arm EVP Drew Henry explains chip design for physical AI prioritizes factors beyond extreme performance and high memory bandwidth. Physical AI is among Arm's most complex computing platforms to build.
钛媒体 · 193d ago
Meta tested its latest AI for content moderation and found it outperforms humans, though only marginally on a previously poor system. Enterprise tools had long detected issues like impossible logins that human moderators missed.
The Register - AI/ML · 193d ago
EK released the EK-Pro GPU WB H200 NVL split water cooling head for Nvidia's H200 NVL PCIe GPU, shrinking it to single-slot FHFL. Dual modules cool GPU cores and HBM separately for optimal flow.
IT之家 · 193d ago
Cursor launched its self-developed AI model surpassing Opus 4.6 on benchmarks. It features a massive price cut, heightening excitement in ambient programming.
量子位 · 193d ago
TeachingCoach is a fine-tuned chatbot providing real-time, pedagogically grounded guidance to higher education instructors. It uses a data-centric pipeline to extract rules from educational resources and generate synthetic dialogues for fine-tuning.
ArXiv AI · 193d ago
Skele-Code offers a natural-language and graph-based interface for non-technical users to build AI agent workflows via interactive notebooks. Steps convert to code, with agents used only for generation and error recovery, avoiding orchestration costs.
ArXiv AI · 193d ago
Researchers propose combining fine-tuning and retrieval for LLM agents to achieve better generalization on unseen tasks. They develop a superior LoRA-based SFT recipe and analyze optimal retrieval strategies for storage, querying, and selection.
ArXiv AI · 193d ago
Researchers developed MultiTraitsss framework using subspace steering to create 'Dark models' that simulate harmful human-AI interactions linked to mental health crises. Evaluations confirm these models produce cumulative harmful behaviors over multi-turn conversations.
ArXiv AI · 193d ago