跟踪前沿趋势，精选高质量内容，帮助你在快速变化的 AI SaaS 世界中保持领先。
How a $900M development center in Warren, Michigan is compressing battery R&D timelines using virtual modeling and small-scale production.
GM is leveraging AI models and a new mini factory to slash EV battery development time in half, aiming to compete with cheaper Chinese imports.
A new GitHub repository provides frozen datasets and scripts to replicate experiments detecting AI-generated text, offering a transparent benchmark for 'AI slop' detection tools.
Data-driven insights reveal how letting forests heal naturally outperforms costly monocultures.
AI analysis of long-term ecological data shows diverse, animal-assisted forests recover faster and store more carbon than traditional tree-planting schemes.
Fields medalist Shing-Tung Yau leads new initiative to integrate AI into mathematical research
Leading Chinese mathematicians have united to form the Yau Laboratory Alliance, aiming to deploy specialized AI models for generating conjectures and formalizing proofs across eight institutions.
Labeling initiatives emerge from UK, France, and US amid literary controversies.
A wave of AI scandals has spurred publishers across Europe and the US to launch certification schemes, aiming to guarantee books are human-authored.
A new YouTube channel called RobynNews features AI presenters that evolve based on viewer comments, orchestrated by a diverse board of models.
Cross-platform desktop app centralizes agent skills for 48 coding tools
Skills Hub v1.0 introduces a unified manager for AI agent skills, letting developers sync tasks across Cursor, Claude Code, and Codex from one desktop app.
Brad Smith argues washing machines face stricter oversight than AI and urges the UK to lead international standards.
Microsoft President Brad Smith contends AI is less regulated than milk, calling for mandatory kill switches and incident reports despite White House resistance.
Plimsoll is a new sandbox service that runs untrusted, agent-authored code while reporting the strength of its isolation boundary, ensuring AI agents can't leak credentials.
Nvidia-backed AI data centre firm cancels listing as investors question pricing.
Firmus has cancelled its planned IPO, valued at over $30bn, citing market volatility and investor concerns that the AI firm's valuation was unsustainable.
Top AI executives are privately planning for a public backlash following a catastrophic AI event, such as a major cyberattack on financial or critical infrastructure.
Easy Fox takes out a loan to keep its LLM-powered driving instructor simulator online.
The developer of 'Teach My Little Sister How To Drive' is spending over $1,000 monthly on AI tokens just to run the demo, raising questions about the viability of LLM-powered games.
Tech titans rebrand AI as 'Super Intelligence' following Trump executive order.
After Trump's directive to replace 'AI' with 'Super Intelligence,' Musk, Benioff, Huang, and Nadella have all publicly aligned with the rebrand.
Executives from Vogue and Esquire are lobbying for the Stealth Bot Prohibition Act to stop disguised AI agents from scraping copyrighted journalism without permission.
The streaming startup launches TV apps, distinguishing itself from AI-saturated rivals with human-produced content and free viewing.
Vurt is expanding vertical video consumption to smart TVs, offering a human-produced alternative to AI-generated microdramas on rival platforms.
WhimKid, an AI storybook generator, turns photos and ideas into illustrated children's books with editable pages and PDF export.
Developer Guksu launches 'suta,' an open-source UI skill for Claude Code and Codex that eliminates generic, repetitive designs by enforcing strict layout and accessibility guidelines.
Jacob Coxon warns that espionage undermines US efforts to control AI pace
Former Anthropic researcher Jacob Coxon asserts he is 90% certain Chinese spies operate within top US AI labs, complicating national security debates.
Microsoft CEO demands 'emergency brake' and mandatory containment for advanced models
Satya Nadella argues we must assume all AI models are compromised from the start, calling for standardized containment and an emergency brake to pause systems mid-task.
Two years after adopting the EU AI Act, officials say Europe is well-equipped to regulate dangerous AI models from US and Chinese firms, with fines up to 7% of global turnover.