Anthropic Mythos AI 太强不敢公开发布
Anthropic 发布 Project Glasswing 网络安全联盟，展示未发布的 Mythos AI 模型。该模型在漏洞检测和编程能力上超越现有所有模型，但 Anthropic 认为其过于强大，仅向 12 家合作伙伴开放，不公开发布。...
latestToday
feed
20 itemslive
Sam Altman 提出 AI 新社会契约
Sam Altman 发表长文阐述 AI 发展与社会契约的关系，呼吁建立新的全球治理框架，强调 AI 安全的重要性。...
Anthropic tells OpenClaw users to pay up
Anthropic blocks third-party agent platforms like OpenClaw from Claude plans, requiring separate API or usage add-on payments. The move addresses unsustainable agent-driven demand. Also covers Granola AI notetaker for phone calls and Netflix's open-source VOID physics-aware video editing framework.
AI 让十亿美元独角兽 solo founder 成为现实
AI 工具正在改变创业游戏规则，让个人创始人能够创建估值数十亿美元的公司。AI 自动化降低了对大型团队的需求。...
Jack Dorsey Makes the AI Case Against Middle Management
Block CEO Jack Dorsey argues AI can replace middle management, framing the company's 40% workforce cut as a bet on the AI era. SpaceX files for the largest IPO in history at a $1.75T+ valuation. OpenAI's "Project Stagecraft" pays 4,000 freelancers to teach ChatGPT specialized job skills.
Dorsey 论证 AI 取代中层管理
Block CEO Jack Dorsey 提出 AI 可取代中层管理的激进观点，作为公司裁员 40% 的理论基础。...
OpenAI Secures $122B Funding to Build AI Superapp
OpenAI raises record-breaking $122B at $852B valuation, with enterprise revenue exceeding 40% and plans to unify ChatGPT, Codex into a "superapp". Anthropic suffers second leak in a week as Claude Code source code goes public. Quinnipiac poll reveals declining American trust in AI, with 70% fearing job losses.
OpenAI's Sora Shutdown: $1M Daily Burn and Disney Blindside
WSJ investigation revealed behind-the-scenes chaos of OpenAI Sora shutdown, including $1M daily burn rate and Disney learning about it less than an hour before announcement, putting a potential $1B partnership on ice.
Anthropic's Secret 'Mythos' Model Leaked: A Step Change in AI Capabilities
Details of Anthropic's next flagship AI model Claude Mythos surfaced after a CMS configuration error, positioning it in a new 'Capybara' tier above Opus. The company confirmed testing a new general-purpose model with advances in reasoning, coding, and cybersecurity.
Exclusive: Alibaba's Kuo Zhang on How Accio Work is Reshaping Enterprise Operations
Alibaba.com President Kuo Zhang sits down with The Rundown AI for an exclusive Q&A on how Accio Work is transforming AI agents from "assistants" to "operators." The interview covers organizational shifts, dynamic workforce deployment, the future of B2B agent ecosystems, and new hiring paradigms—revealing that a one-person billion-dollar company may be just months away.
Meta Open-Sources TRIBE v2: Brain AI That Outperforms Real Scans
Meta open-sourced TRIBE v2, a brain simulation AI trained on 700+ subjects whose synthetic predictions outperform real fMRI scans; Apple plans to open Siri to rival AI assistants in iOS 27; Wikipedia voted 40-2 to ban AI from writing articles.
ARC-AGI-3 Resets Frontier AI Scoreboard
François Chollet released ARC-AGI-3 benchmark where humans solve 100% on first try but top AI models score below 1%. Reddit CEO outlined human-bot separation plan. Google introduced TurboQuant algorithm compressing AI memory 6x.
OpenAI Winds Down Sora to Prioritize Spud Model
OpenAI shut down Sora video generator to redirect compute to upcoming Spud model. Figure AI founder launched AI device startup Hark. Apple testing standalone Siri app and iOS 27 chatbot experience.