AI Weekly Digest #5· Dec 22–28, 2025 · AI Factories, Deceptive Models & the Agentic Economy скачать в хорошем качестве

AI Weekly Digest #5· Dec 22–28, 2025 · AI Factories, Deceptive Models & the Agentic Economy

🤖 Welcome to AI Weekly Digest — your concise briefing on how AI is turning into global infrastructure. Main body: This week in AI, the headlines were less about new apps and more about power, ownership and control. Nvidia deepened its ties with Groq, SoftBank raced to deliver a $22.5B commitment to OpenAI, and Japan placed a multi-billion bet on “physical AI” and humanoid robots. At the same time, reporters and authors launched fresh lawsuits over training data, and China drafted new rules for human-like AI companions. On the model side, DeepSeek-V3 and GLM-4.7 pushed open-source forward, VL-JEPA showed how tiny vision-language models can match giants, and new tools from Qwen, Google and Nvidia quietly upgraded the AI toolbox. This episode connects the dots between: • AI as heavy infrastructure (chips, data centers, physical robots) • The “original sin” of AI — training on copyrighted works • Regulations for human-like AI systems in China and beyond • New efficient models and research that will shape 2026 ▶️ Watch the full breakdown, then answer the Weekly Question in the comments. Chapters: 00:00 Intro — AI becomes infrastructure 00:47 Segment 1 — Money, Chips & Data Centers 01:12 Segment 2 — Rules, Risks & Pushback 02:07 Segment 3 — Models, Research & Everyday AI 05:05 Segment 4 — Weekly Takeaways & Big Question Weekly Question: Who should decide how close AI is allowed to get to us when it looks and talks like a human — app stores, regulators, or the platforms that own the infrastructure? Disclosure: This video is for informational purposes only. We are not affiliated with or sponsored by any company mentioned. #ainews #AINews #ArtificialIntelligence #TechNews #AIWeekly #FutureOfAI #AIShortsDigest