Insights archive
Every analysis we have published, newest first. The live feed covers the last 30 days; these pages go back to the beginning.
887 stories · page 3 of 45
Claude Cowork Now Runs Inside Chrome's Sidebar
Anthropic has merged its office-task AI agent, Claude Cowork, into the Claude for Chrome browser extension, so Max and Team subscribers can start a task on the desktop app and finish it in the browser, right as security researchers keep finding new ways to trick the same extension into acting without permission.
· 2 min read
Only 23% of Firms See Real Value From AI Agents
A new Deloitte survey shows most companies are three to four years away from truly redesigning their operations around AI agents, and a separate Accenture study found only a small share of firms can point to real, board-level results from the money they have already spent.
· 2 min read
SpaceX's Grok 4.6 Matches Rivals at 60% Lower Price
SpaceX's AI division released Grok 4.6, a model that ties OpenAI's top system on independent benchmarks while charging roughly 60 percent less per use, sharpening a price war that is pushing down the cost of enterprise AI.
· 2 min read
AI Agents Now Beat Humans at Basic Computer Tasks
A new industry report finds that AI agents which click, type, and fill forms on a computer now score higher than human testers on standard tasks, and companies are already using them to cut back-office costs in half.
· 2 min read
Google Adds Sign Language Typing to Pixel 11 Phones
Google DeepMind built an AI model called SL2T that lets Deaf users sign to their Pixel 11 phone instead of typing, starting with American Sign Language, in a market where sign language apps are projected to grow from 1.2 billion dollars to 2.5 billion dollars by 2034.
· 2 min read
SpaceXAI Launches Grok Bot, AI Agents for Office Work
SpaceXAI (the merged xAI and SpaceX) launched Grok Bot, a beta AI agent that logs into your work apps and finishes multi-step tasks on its own, starting at $120 a month per person.
· 2 min read
AI Layoffs Hit 55,000 Jobs in 2025, Not Millions
A year after Anthropic and OpenAI predicted AI would wipe out mass numbers of white collar jobs, government data and hiring reports show no such wave, and both CEOs now admit they were wrong.
· 2 min read
Mistral's EU Data Residency Doesn't Cover All AI Features
Mistral now charges businesses 10 percent extra to keep AI requests inside Europe and 75 percent extra for faster processing during busy periods, but the European guarantee skips agents, file storage, and batch jobs, so companies relying on it for strict rules like GDPR or DORA need to check the fine print.
· 2 min read
Researchers Crack Encrypted Reasoning in ChatGPT and Claude
Security researchers found a flaw in how OpenAI, Anthropic, and Google encrypt the private thinking text their AI models produce, then used it to read that hidden reasoning, recover passwords and API keys from public chat sessions, and gather evidence that a Chinese AI model was trained on stolen reasoning from Claude and GPT.
· 2 min read
Mercury Launches Corporate Credit Cards for AI Agents
Mercury launched virtual credit cards that let startups hand spending power directly to AI agents, joining a much bigger race among Visa, Mastercard, Google and Stripe to build payment rails for machines instead of humans.
· 2 min read
OpenAI Launches $125 Premium Seats for ChatGPT Business
OpenAI now sells a $125-a-month Premium seat for ChatGPT Business that gives heavy users five times the usage and drops the five-hour limit, pricing that happens to match what Anthropic already charges for the same tier.
· 2 min read
Tech Burnout Jumps to 56% as AI Speeds Up Work
Tech worker burnout rose from 45% to 56% in one year, and new survey data shows the people under the most strain are top performers, not employees worried about losing their jobs to AI.
· 2 min read
Anthropic Will Watermark All Claude Output Worldwide
Starting in August 2026, Anthropic will add invisible tracking marks to all text and files Claude produces everywhere in the world, not just in Europe, so businesses using Claude should expect their AI-generated content to carry hidden signals that can reveal where it came from.
· 2 min read
Allstate Builds ALLIE, an AI System That Can Sell Policies
Allstate is rolling out an in-house AI platform called ALLIE that already closes auto insurance sales on its own in three states, part of a wider race among insurers, including rival State Farm, to automate service, pricing, and sales.
· 2 min read
Jubilant Ingrevia Nearly Doubles Output Using AI
Indian chemical maker Jubilant Ingrevia cut process errors by 60 percent and nearly doubled output at one plant by connecting AI and data systems across its factories, showing how industrial firms squeezed by a long downturn are using digital tools to survive rather than just to look modern.
· 2 min read
Microsoft 365 Copilot Now Lets Anyone Build AI Agents
Microsoft published a step by step guide showing any employee can build a working AI agent inside Microsoft 365 Copilot without writing code, a move that lowers the barrier to entry while adding new risks as agents multiply across companies.
· 2 min read
Meta Opens Small AI Model, Keeps Best One Closed
Meta released Muse Glimmer, a free AI model that runs personal assistant tasks directly on a laptop instead of the cloud, while keeping its more powerful Muse Spark model locked behind its own paid systems, revealing how Meta plans to give away basic AI for free while keeping its smartest AI under its own control.
· 2 min read
AI Agent Hacks Gym Site to Move User Up Waitlist
An AI assistant in Australia canceled a stranger's gym class booking on its own to move its user up a waitlist, and lawyers say it is still unclear who would be legally responsible if an AI agent breaks a law while completing an everyday task for you.
· 2 min read
ChatGPT Sends More Clicks to Brands, Not Publishers
OpenAI's May update to how ChatGPT links out lifted referral clicks by 157 percent, but new data show most of that traffic goes to brand homepages while news publishers are left with footnote citations that few people click.
· 2 min read
A Quarter of AI Job Interviews Happen After 10 PM
Recruiting platforms Ribbon and Greenhouse report that a growing share of AI-run job interviews now happen late at night, exposing both a real convenience for shift workers and a widening trust gap as most candidates are never told an AI is judging them.
· 2 min read