Welcome to today’s roundup
AI adoption is surging on every front. Google Search’s AI summaries now reach two billion people each month while OpenAI is pouring gigawatts of new compute into a giant Texas data‑center project. DeepMind has an ancient‑history helper that can date and patch broken Latin inscriptions, Google is seeding AI video tools in Photos and YouTube Shorts, GitHub is giving every open‑source dev a free inference API, and Proton has launched a fully encrypted chatbot for privacy‑minded users. Read on for the details.
AI You’ll Actually Understand
Cut through the noise. The AI Report makes AI clear, practical, and useful—without needing a technical background.
Join 400,000+ professionals mastering AI in minutes a day.
Stay informed. Stay ahead.
No fluff—just results.
1. Google AI Overviews tops two billion users
Alphabet CEO Sundar Pichai told investors that the AI Overviews feature, which puts a generated summary at the top of many Google Search results, now logs two billion monthly users across two hundred countries. Gemini’s standalone app has grown to 450 million monthly users and the chat‑style AI Mode inside Search already counts more than 100 million users in the United States and India. Google says Deep Search and more personalized answers are “coming soon,” and notes that developers have issued nine million Gemini API keys while public Veo‑powered tools have produced over seventy million videos since May.
2. OpenAI and Oracle add 4.5 gigawatts to the Stargate project
OpenAI confirmed a new agreement with Oracle that will build 4.5 GW of additional data‑center capacity in Abilene, Texas, pushing the total under development past five gigawatts and powering more than two million AI chips. The expansion is part of OpenAI’s promise to invest 10 GW of U.S. infrastructure worth roughly 500 billion dollars within four years and is expected to create about 100,000 construction and operations jobs. The company says the first racks of Nvidia GB200 GPUs are already online and training next‑gen frontier models.
3. DeepMind’s Aeneas helps historians read lost Roman inscriptions
The Aeneas system was trained on nearly 200,000 Latin inscriptions, totaling sixteen million characters, so it can guess missing words, pinpoint an inscription’s province among sixty‑two Roman territories, and date a fragment to within about thirteen years. In testing with twenty‑three scholars, Aeneas provided useful context nine times out of ten and even offered credible dates for the Res Gestae Divi Augusti imperial monument. Researchers say the model could democratize epigraphy by replacing years of manual cross‑referencing with instant suggestions.
4. AI video tools reach Google Photos and YouTube Shorts
Starting today, U.S. users of Google Photos on Android and iOS can turn a single picture into a six‑second AI video clip by choosing prompts like “Subtle movements,” while YouTube Shorts creators in the U.S., Canada, Australia and New Zealand get a similar generator with flexible clip lengths. All outputs carry SynthID watermarks, and Photos will soon add a Remix feature that converts images into styles such as anime or 3‑D animation inside a new Create tab. Google warns results are experimental and invites thumbs‑up or thumbs‑down feedback.
5. GitHub Models offers a free inference endpoint for open source
GitHub’s new Models service gives every user a no‑cost, OpenAI‑compatible API that hosts popular models like GPT‑4o and Llama 3. Open‑source maintainers can swap a single baseURL in their code so contributors no longer need paid keys or heavyweight local models. The free tier works in any environment, including GitHub Actions, while a metered paid tier unlocks higher rate limits and 128‑k‑token contexts. GitHub says the goal is to remove the “last mile” barrier that keeps hobbyists from trying AI‑powered tools.
6. Proton launches Lumo, an end‑to‑end encrypted chatbot
Privacy‑focused Proton, maker of Proton Mail and Proton VPN, has introduced Lumo, a chat assistant that secures every conversation with zero‑access encryption so only the user’s device can read the content. Lumo runs on open‑source models hosted on Proton’s European servers, defaults to privacy‑friendly search, and will analyze files stored in Proton Drive without leaving encrypted storage. The free plan allows up to one hundred interactions a week, and the twelve‑dollar‑ninety‑nine‑cent Lumo Plus tier removes that cap and adds advanced features.
Turn AI from Party Trick to Profit Machine
I built Infinity for one goal: hand you the same AI systems and automations I lean on every day.
Plug them in, cut the busy work, and watch new revenue pop up.
You’ll quickly become the go to AI expert everyone trusts.
Ready to plug in?
How would you rate today's newsletter?
Stay tuned for more updates, and have a fantastic day!
Cheers,
Zephyr