OpenAI just rolled out Record mode on ChatGPT's macOS desktop app for Pro, Enterprise, and Edu users. You can record meetings and brainstorming sessions directly in the app, and ChatGPT handles the transcription, summarizes key points, and drafts action items or code snippets. It basically turns ChatGPT into a built-in meeting assistant.
Google just launched Search Live in its mobile app, letting users have free-flowing voice conversations with Search powered by a custom Gemini model. It streams spoken answers, shows web links on-screen, and keeps running in the background. It basically turns search into an active audio call, though it's currently limited to U.S. Labs users.
A UCLA graduate just went viral after proudly opening his laptop on stage during commencement to reveal an active ChatGPT window, crediting the AI for helping him finish his final projects. The clip racked up over 43 million views on X, sparking a heated online debate over whether this bold move is modern efficiency or an academic integrity red flag.
Google just released Gemini 2.5 Pro and Flash as stable models while introducing a preview of 2.5 Flash-Lite. Flash-Lite is pitched as the fastest and most cost-effective 2.5 model yet, packing a 1M-token context window, adjustable thinking budgets, and lower latency for high-volume tasks. Why it matters: Google is aggressively undercutting API costs for high-throughput production workloads.
An advertising watchdog just told Microsoft to drop or modify claims that Microsoft 365 Copilot boosts productivity and ROI. Regulators found surveys claiming up to 75% of users "feel" more productive reflect subjective perception rather than objective proof, while confusing branding around Business Chat misleads customers. Microsoft disagreed with the verdict, but agreed to clarify its marketing.
OpenAI just notified developers that GPT-4.5 Preview will be removed from its API on July 14, 2025. While the phase-out was technically planned back in April alongside GPT-4.1's debut, the reality check has left teams relying on its creative writing tone scrambling to migrate. ChatGPT subscribers can still breathe easy: it remains an option in the web dropdown.
MiniMax just open-sourced MiniMax-M1, a hybrid-attention reasoning model with a 1 million token context window and 80k-token reasoning output. Thanks to its Lightning Attention mechanism and CISPO RL algorithm, deep reasoning costs roughly 30% of DeepSeek R1's compute, with the entire RL phase trained on 512 H800s for just $535k. It rivals Gemini 2.5 Pro in long context while being completely free on the web.
Adobe just rolled out Adobe LLM Optimizer at Cannes Lions, an enterprise tool designed to help brands rank better in ChatGPT, Claude, and Gemini responses. It tracks how models cite brand content and automatically suggests or applies structural and metadata fixes. Basically, Adobe is making Generation Engine Optimization (GEO) the next SEO.
Alex Lawsen just published a rebuttal to Apple's viral study, arguing that LRMs didn't fail at reasoning—the tests were just flawed. Models hit token generation ceilings rather than brain freezes, and even refused mathematically impossible river-crossing puzzles. When asked to write code instead of typing out moves, Claude and o3 solved 15-disk Hanoi puzzles easily.
Google just launched Audio Overviews in Search Labs, using its latest Gemini models to turn search queries into quick, conversational audio recaps. The built-in player also serves up relevant links directly on the results page if you want to dig deeper. It basically brings the NotebookLM audio magic directly into everyday Google searches.
OpenAI just supercharged ChatGPT Projects with support for Deep Research and Voice Mode. The update also improves project memory so it can pull context from past conversations, while mobile users can now upload files and switch models directly in the app.
Microsoft just rolled out Copilot Vision with Highlights for Windows 10 and 11 users in the US for free. Clicking the new glasses icon lets the AI see your selected app, talk through tasks in real time, and highlight exactly where to click next. It is opt-in, doesn't require Copilot Pro, and remains limited to US users for now.
Meta just rolled out generative AI video editing across the Meta AI app, Meta.AI website, and Edits app. Users in the US and over a dozen countries can choose from over 50 preset prompts to transform 10 seconds of video into comic styles or video games for free. Meta says the tool was inspired by its Movie Gen research, with custom text prompts coming later this year.
The Browser Company just launched Dia, a new AI-first browser now in beta for existing Mac Arc users. Dia returns to a familiar Chrome-like layout but embeds an AI assistant directly into the sidebar that can read your open tabs, cookies, and browsing history. Why it matters: it bets that your browser should be your primary everyday AI companion.
Starbucks is piloting an AI virtual assistant called Green Dot Assist across 35 stores to support baristas in real time. Powered by Microsoft Azure's OpenAI platform, the in-store iPad tool answers recipe questions, troubleshoots espresso machines, and helps managers cover shifts. A wider rollout across the U.S. and Canada is planned for fiscal 2026.
OpenAI just rolled out o3-pro, an upgraded reasoning model that replaces o1-pro for ChatGPT Pro and Team subscribers, as well as API developers. It brings built-in tool access like web browsing and Python execution, beating rivals on tough math and science tests, though it takes longer to respond and skips image generation for now.
Snap just announced plans to sell its lightweight AR glasses, Specs, to everyday consumers starting in 2026. The new see-through frames pack an onboard multimodal AI assistant to process video and audio, alongside developer support for OpenAI and Google DeepMind models.
MIT researchers just monitored brain activity during essay writing and found that ChatGPT users had significantly weaker neural connectivity than those using search engines or no tools at all. When forced to write unaided later, former AI users showed lower brain engagement and struggled to recall their own work. Why it matters: while AI tools save immediate time, heavy reliance might quietly hollow out critical thinking skills.
Google's AI Overviews and new chat-driven search tools are devastating referral traffic for digital publishers, according to a Wall Street Journal report. Users now get quick answers directly from chatbots instead of clicking blue links, sending The New York Times' organic search traffic share down to 36.5% in April. To survive, media outlets are rushing into licensing and revenue-sharing deals with AI giants.
Mistral AI just introduced Magistral, its first-ever reasoning model family designed for transparent, multi-step problem solving. It comes in two flavors: an open-weight 24B model (Magistral Small) and an enterprise version (Magistral Medium). Why it matters: it reasons natively across multiple languages and claims up to 10x faster throughput via Flash Answers in Le Chat.
