Meta just cut 8,000 jobs while pouring $125B into AI, Google's I/O afterglow brings real tools for classrooms, and India is writing AI policy state by state.
Gemini 3.5 Flash is now the AI inside Google Search, for everyone
Following Google I/O, Gemini 3.5 Flash is now generally available and set as the default model in AI Mode in Search globally. Google says it matches larger flagship models in quality while running at Flash speeds. Google Antigravity 2.0, a desktop app for running and coordinating multiple AI agents, also launched this week alongside the new AI-reimagined Search box, described as the biggest update to the Google Search interface in 25 years.
Why it matters: This is the AI most people will interact with without ever choosing to, it's simply inside the tool they already use. If Google Search changes how you find information, this is the moment that started.
Meta cuts 8,000 jobs and moves 7,000 others onto AI teams
Meta began implementing the 8,000 layoffs announced this month (about 10% of its global workforce), while simultaneously reassigning more than 7,000 employees to newly created AI-focused roles covering products, assistants, and automated agents. The company projects capital expenditure of $125–$145 billion for 2026, almost entirely directed at AI infrastructure.
Why it matters: This is the clearest illustration yet that Big Tech's AI investment is being funded, in part, by headcount reductions. AI fluency is no longer an advantage, it is becoming a survival requirement for people working in or adjacent to tech.
Anthropic upgrades Opus and launches Claude Design
Anthropic announced upgrades to its Opus class of models, with stronger performance on coding, professional tasks, and multi-step agentic workflows. Separately, Anthropic Labs released Claude Design, a tool that lets users collaborate with Claude to create polished, production-ready visual designs. The company is also in early talks with Microsoft about running Claude inference workloads on Microsoft's custom Maia 200 AI chips via Azure.
Why it matters: Anthropic is competing hard on both capability and product. Claude Design is a direct play for the creative workflow market, and the Microsoft chip discussion signals deeper infrastructure integration is coming.
Google commits $150M+ to train 6 million educators in AI
Google's AI Educator Series launched this week, offering free, comprehensive AI training to all 6 million K–12 and higher education teachers in the US, with new modules releasing monthly. Boston Public Schools also announced it will become the first major-city district to make AI fluency a high school graduation requirement starting September 2026, backed by a $1 million seed grant. The State University of New York added AI literacy to its general education core for all incoming undergraduates.
Why it matters: The US education system is beginning to treat AI literacy the same way it treats reading or maths, as a baseline competency. Students starting school now will graduate into a world where this is table stakes.
Rajasthan launches AI-ML Policy, India's data centre capacity to grow 4–5x by 2030
Rajasthan unveiled its AI-ML Policy 2026 at a regional AI impact conference in Jaipur, joining a growing list of Indian states formalising AI governance. At the national level, India's data centre capacity currently stands at 1,280 MW and is forecast to grow four to five times by 2030, driven by cloud adoption and AI workloads. India Semiconductor Mission 2.0, announced in the Union Budget, allocates ₹1,000 crore for chip research, training, and workforce development. The AIKosh public dataset platform now hosts 7,500+ datasets and 273 AI models across 20 sectors.
Why it matters: India is building AI infrastructure from the bottom up, hardware, data, policy, and talent pipelines at the same time. The states writing policy now will shape where the opportunities land.
Trump administration softens on AI oversight after "Mythos" national security concern
Reports indicate the Trump administration is actively reconsidering its earlier resistance to AI oversight, prompted by concerns about Anthropic's "Mythos" model, described as capable of autonomously identifying and exploiting cybersecurity vulnerabilities. The shift marks one of the clearest examples of national security pressure pushing AI regulation faster than any policy advocate could. Separately, the 2026 International AI Safety Report flagged deepfake proliferation and autonomous cyber-exploitation as the most urgent near-term risks.
Why it matters: When a government that opposed AI rules starts drafting them, the direction of the global regulatory conversation changes. Other governments will treat this as a signal.
AMD's 2nm chip production has started, the next AI hardware race is live
AMD has begun production of its 6th Generation EPYC "Venice" processors on TSMC's 2nm process node, the first major high-performance computing product to enter production at this scale. AI capability is, at its foundation, a hardware race. Every model upgrade announced this week runs on chips; the chips arriving next year will make today's launches look like early drafts. Watch who controls this layer, because it will determine who controls the models above it.
Gemini Spark launches next week + Cursor hits $29B + SUNY mandates AI literacy
Google's Gemini Spark personal agent (which acts on your behalf across your digital life) arrives next week for Google AI Ultra subscribers in the US. Anysphere, the company behind the Cursor AI coding editor, closed a $2.3 billion Series D at a $29.3 billion valuation, reflecting how much demand there is for AI-assisted software development. And the State University of New York quietly made AI literacy part of general education for every incoming undergraduate starting this fall.