Three rival AI models, a landmark protocol update, $14B in AI infrastructure, and India shaping global AI rules: the kind of brief that proves we're not slowing down.
Three AI giants launched this month, and there's now a clear #1
Claude Opus 5 topped the Agentic AI Index at 55.3 points this week, nudging past OpenAI's new GPT-5.6 Sol (54.0) and xAI's Grok 4.5. OpenAI launched its GPT-5.6 family in three sizes (Luna, Terra, and Sol), while xAI's Grok 4.5 arrived with a 500,000-token context window and strong coding benchmarks. All three shipped within weeks of each other in July, making this the most competitive month in AI model history.
Why it matters: Whether you use Claude, ChatGPT, or Grok, every tool you rely on just got meaningfully smarter. The rivalry drives speed: features that used to take months now appear in weeks. Picking a favourite matters less than learning to use whichever one fits your work.
The rulebook for AI agents just got its biggest update since day one
The Agentic AI Foundation (a Linux Foundation project) released the MCP 2026-07-28 specification yesterday, described as the largest change to the Model Context Protocol since its launch. The update moves from a persistent, stateful connection to a simpler request-and-response model, letting AI agents run on cheaper serverless and edge infrastructure. MCP already has 400 million monthly SDK downloads and is the de facto standard for connecting AI agents to apps and data.
Why it matters: MCP is the common language AI agents use to connect to your files, tools, and apps. A simpler protocol means more developers can build agents, which means more of your daily work could get automated, sooner than you might expect.
Meta and BlackRock are spending $14 billion on a single AI campus in Texas
Meta and BlackRock (alongside Global Infrastructure Partners and HPS Investment Partners) announced a joint $14 billion plan to build a 1 gigawatt AI data centre campus in El Paso, Texas. The campus won't come online until 2028. It is one of the largest single AI infrastructure commitments ever announced.
Why it matters: Data centres are the factories of the AI era. A $14 billion bet on one campus signals that the biggest companies believe demand for AI compute will keep growing for years, and that today's tools are just the beginning.
Google training 6 million teachers, and New York's 64-campus university just made AI literacy mandatory
Google is rolling out free AI literacy training to 6 million K-12 and higher education educators in the US through a partnership with ISTE+ASCD, including career certificates and access to Gemini tools. Separately, the State University of New York (64 campuses, hundreds of thousands of students) voted to embed AI literacy into general education requirements for every incoming undergraduate from Fall 2026. NotebookLM access has also been doubled for educators on Google's Education plans.
Why it matters: When the largest public university system in the US makes AI literacy a graduation requirement, it redefines what baseline employability looks like. If you're a student or a teacher anywhere in the world, this is the signal you've been waiting for.
AI deepfakes are now a legal weapon, and job interviews are the new frontline
The US DEFIANCE Act creates a federal right of action for victims of non-consensual AI deepfakes, allowing them to sue creators, distributors, and platforms that host the content. On the employment side, Gartner is projecting that 1 in 4 job candidate profiles globally will be AI-generated by 2028. Experian's 2026 Fraud Forecast identifies synthetic job candidates that can pass live video interviews in real time as one of the top emerging threats.
Why it matters: Deepfakes aren't just a celebrity or politics problem. They're entering hiring, HR, and everyday legal life. Companies and individuals both have very few defences yet, and policymakers are only just catching up.
India shaped the global AI governance agenda, and its GPU count just tripled
The India AI Impact Summit 2026, held in New Delhi, produced the "New Delhi Frontier AI Impact Commitments", a set of responsible AI pledges submitted to Rajya Sabha by the IT Minister last week. The IndiaAI Mission has now provisioned 38,000 GPUs (up from the original 10,000 target), with more on the way. Indian AI startups also raised $431 million in a single week this July, led by Emergent and Reo.Dev: though overall July funding has cooled from June's record.
Why it matters: India isn't just participating in the global AI conversation, it's helping write the governance rulebook. For students and professionals in India, the infrastructure, the funding, and the policy window are all open at the same time. That combination doesn't last forever.
A company building AI that improves itself just signed a $400M deal
Recursive Superintelligence (founded by Richard Socher, former chief scientist of Salesforce) signed a $400 million multiyear compute contract with AWS. The company's stated goal is to build "open-ended self-improving systems," meaning AI that rewrites and upgrades itself rather than waiting for humans to release new versions. Initial products are expected by October 2026. This is a genuinely different category from a chatbot or a coding tool. Keep it in your peripheral vision.
Claude in your glasses, 169 US AI laws, and India's funding pace shifts
Innovative Eyewear added Claude AI to its entire Lucyd smart eyewear lineup, free for all customers: AI is now literally on your face. US states have passed 169 AI laws since 2022 (with 146 more bills pending), creating a patchwork that's hard for any business to track. And India's overall startup funding in July 2026 has pulled back to roughly $820M after a record $1.9B in June: AI is still the leading sector, but investors are spreading across manufacturing, healthtech, and deep tech.