https://www.winzheng.com/en/article/mythos-5-aisi-github-supply-chain-attack-student-report Winzheng en 2026-08-24T19:00:25+08:00 Mythos 5 Launches Real GitHub Supply Chain Attack in AISI Evaluation; Student Report Exposes AI Deception AI安全, 供应链攻击, Anthropic Mythos 5 https://www.winzheng.com/en/article/children-outlearn-ai-mystery Winzheng en 2026-08-24T18:24:45+08:00 Kids outlearn AI—and we still don’t know why Child Language Learning, AI Language Models, Language Acquisition, Cognitive Science, Artificial Intelligence https://www.winzheng.com/en/article/teachers-deepfake-epidemic-schools Winzheng en 2026-08-24T18:24:15+08:00 They Dedicated Their Lives to Teaching. Then the Deepfakes Started Deepfake, AI Safety, School Violence, Teacher Rights, Women Safety https://www.winzheng.com/en/article/anthropic-aug-2026-risk-report-benchmark-saturation-model-2- Winzheng en 2026-08-24T14:20:16+08:00 Safety Benchmarks Saturate Before Capabilities: Anthropic Risk Report Reveals Structural Cracks in Self-Certification Anthropic, AI安全, 模型对齐, 风险报告, Mythos 5, Model 2, 基准饱和, 多智能体 https://www.winzheng.com/en/article/alibaba-hk-share-placement-10-2bn-ai-expansion Winzheng en 2026-08-24T06:13:18+08:00 Alibaba Raises US$10.2 Billion in Hong Kong Share Placement, Fully Allocated to Full-Stack AI Capability Building 阿里巴巴, AI基础设施, 香港配售, 云服务 https://www.winzheng.com/en/article/eu-ai-act-enforcement-august-2026-prohibition-vs-high-risk-d Winzheng en 2026-08-24T06:09:55+08:00 EU AI Act Enforcement Begins: Bans Take Effect Immediately, While High-Risk Compliance Quietly Delayed to End of 2027 欧盟AI法案, AI监管, 合规, GPAI, 数字综合法, AI治理 https://www.winzheng.com/en/article/linkdaze-smart-calendar-family-run Winzheng en 2026-08-24T04:24:30+08:00 Linkdaze’s smart calendar is built to run a household, not just track a schedule Smart Calendar, Family Management, AI Tools, Freemium, Tech News https://www.winzheng.com/en/article/mysterious-ai-model-ox-alpha Winzheng en 2026-08-24T04:23:47+08:00 Who’s behind the new ‘stealth model’ Ox Alpha? AI Model, Ox Alpha, Tech Mystery, Social Media, AI Safety https://www.winzheng.com/en/article/ai-training-copyright-books-legal Winzheng en 2026-08-24T00:24:16+08:00 Is it legal to train AI models on copyrighted books? It’s complicated AI Copyright, Fair Use, LLM Training, Tech Law, Authors' Rights https://www.winzheng.com/en/article/flock-ceo-compromise-surveillance-backlash Winzheng en 2026-08-24T00:23:47+08:00 Flock CEO calls for ‘compromise’ as surveillance company faces growing backlash AI Surveillance, Privacy, Flock Safety, Tech Ethics, Social Controversy https://www.winzheng.com/en/article/ai-agents-surpass-humans-token-usage-14x-openrouter-2026 Winzheng en 2026-08-23T20:20:01+08:00 AI Agent Compute Consumption Jumps 14x in Six Months, Overtaking Human Users as Largest Inference Market Buyer AI代理, 推理算力, OpenRouter, 代理经济, 大模型成本, DeepSeek, 基础设施 https://www.winzheng.com/en/article/google-a2a-protocol-agentic-ai-foundation-transfer Winzheng en 2026-08-23T14:22:32+08:00 Google A2A Protocol Transferred to Agentic AI Foundation, Now Jointly Governed by 250+ Members AI协议, Agent互操作, 中立治理 https://www.winzheng.com/en/article/claude-protein-design-14-15-targets-wet-lab-biosafety Winzheng en 2026-08-23T14:18:45+08:00 Claude Protein Design Hits 14/15 Targets: First Wet Lab Validation Stirs Biosecurity Governance Debate 蛋白质设计, Anthropic, 生物安全, AI科学, Claude, 双重用途风险, 药物研发 https://www.winzheng.com/en/article/harvard-ai-avatar-startup-bootcamp Winzheng en 2026-08-23T06:23:39+08:00 Harvard’s $699 startup bootcamp offers AI avatars of its instructors AI education, Startup bootcamp, Harvard Business School, AI avatars https://www.winzheng.com/en/article/openai-chatgpt-for-teens-launch-analysis Winzheng en 2026-08-23T06:11:36+08:00 OpenAI Launches ChatGPT for Teens: Safety Tools and Cognitive Risk Concerns Coexist OpenAI, ChatGPT for Teens, 青少年AI, 产品战略, 内容安全 https://www.winzheng.com/en/article/anthropic-ai-agent-kills-rivals-simulation-alignment-warning Winzheng en 2026-08-23T06:08:29+08:00 AI Agents Kill Their Own Kind and Refuse to Rescue People in Simulations: Anthropic Sounds Misalignment Alarm AI安全, Anthropic, AI对齐, AI代理, 风险评估 https://www.winzheng.com/en/article/inherent-faraday-ai-research-replication Winzheng en 2026-08-23T04:25:22+08:00 Inherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research AI Agent, Faraday, Research Automation, DeepMind, AI for Science https://www.winzheng.com/en/article/openai-california-ai-safety-bill Winzheng en 2026-08-23T02:23:56+08:00 OpenAI says California should strengthen its AI safety bill AI Safety, OpenAI, California Legislation, Regulatory Strategy, Tech Companies https://www.winzheng.com/en/article/frontier-ai-rogue-model-containment Winzheng en 2026-08-23T00:23:52+08:00 Frontier AI labs still won’t say how they’d contain a rogue model AI safety, rogue model, frontier AI, AI regulation https://www.winzheng.com/en/article/anthropic-mythos-5-claude-security-enterprise-35m-fund Winzheng en 2026-08-22T20:25:23+08:00 Anthropic Brings Mythos 5 to Claude Security Enterprise and Establishes $35 Million Fund AI安全, Anthropic, Claude Security, 企业工具 https://www.winzheng.com/en/article/nvidia-poolside-6b-model-factory-ai-infrastructure-control Winzheng en 2026-08-22T20:21:10+08:00 NVIDIA's $6 Billion Poolside Model Factory License: Computing Power Giant's Third Move Targets the Full AI Stack 英伟达, Poolside, AI基础设施, 模型评测, 反垄断, 授权协议, 算力, AI投资