The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
Meta has bought Conductor, a tool for running several AI coding helpers at once. Its cloud service shuts down November 20, so users have a deadline.
Read more: Meta buys Conductor, a tool that lets one person manage many AI codersHouse Republican leaders want Nvidia's CEO to brief them privately, after a report on his ties to an executive linked to Chinese intelligence. Nvidia chips power most AI.
Read more: House Republican leaders ask Nvidia's CEO to brief them privatelyOpenAI's new ChatGPT Slides builds decks inside the chatbot, but reports say Japanese text can go wrong when you export to PDF.
Read more: ChatGPT can now build slide decks, but Japanese PDFs can breakA new test of AI agents on simulated professional work found none scored above 12 percent, a reality check for anyone planning to hand real jobs to AI.
Read more: Top AI agents finish barely 12 percent of real office-style workCloudflare warns that AI agents acting for millions of people look like one visitor to websites, which makes fraud, disputes and blocking harder to handle.
Read more: AI assistants shopping for you leave websites unable to tell who is whoA user says Claude Code put his email address in web requests during searches, and that he now gets dozens of scam emails a day. Anthropic has not replied in the report.
Read more: User says Claude Code shared his email address with websitesA new app called Oraku says it can turn a photo or one sentence into a short video in well under a minute, and it shows the credit cost before you start.
Read more: New App Oraku Turns a Photo or a Sentence Into a VideoDisney tech staff asked leaders when the next layoffs would come, and say they got no clear answer. The episode shows what silence does to a workforce.
Read more: Disney tech staff press bosses on layoffs they read about in the newsAirbnb says its own AI travel agent arrives in the first half of 2027, planning trips and handling refunds and rebookings inside the app.
Read more: Airbnb plans an AI travel assistant for the first half of 2027Perplexity has launched Teammates, AI helpers that work around the clock on business tasks. They require its $200 a month Max plan.
Read more: Perplexity launches AI helpers that work around the clockOpen AI models that companies can run themselves now beat or match paid rivals on some finance tasks, which could mean lower costs and more control over private data.
Read more: Free AI models from Kimi and Mistral now rival the top paid onesNvidia's RTX Spark laptops and desktops can run several AI tools at once on the device itself, which Microsoft's Copilot+ PCs never really managed.
Read more: Nvidia's new RTX Spark computers run real AI without the internetMETR says Anthropic hid much of its best evidence from the public version of a 100-page AI assessment, so readers must trust METR's word more.
Read more: Outside AI tester says Anthropic blacked out key evidence in its reportAds are now inside AI chatbots, and one ad firm says it often sees a user's whole conversation before it picks which ad to show.
Read more: Ads Are Moving Into AI Chatbots, and They Can Read Your ChatsA now-fixed flaw let attackers get around ChatGPT's Lockdown Mode and possibly pull chat data out without the user clicking anything.
Read more: Researchers find a way around ChatGPT's strictest privacy settingFake OpenAI, SpaceX and Anthropic event pages are tricking employees into handing over Microsoft and Google work sessions, even with two-step login on.
Read more: Fake AI and SpaceX event invites are stealing work loginsA man is suing OpenAI, claiming ChatGPT helped draft a poisoning plan against him. The case could shape who is responsible when a chatbot gives harmful answers.
Read more: Man sues OpenAI, says ChatGPT helped plan his poisoningGoogle is building tools so Gemini can finish multi-step jobs across your Android apps, with some of the work staying on your phone.
Read more: Google lets AI assistants carry out tasks across Android appsAn Android Police writer uses Gemini by voice for email, daily planning and chat catch-up. Here is what works, and one limit on WhatsApp.
Read more: Six ways to use your voice with Gemini to get work done on AndroidGoogle Forms is getting a Gemini AI feature that reads survey answers and flags trends, which could save hours for anyone who collects feedback.
Read more: Google Forms will use Gemini AI to summarize your survey answersOpenAI's CEO said it was terrible for everyone to own the world's scarce memory chips, then reversed course after online anger. Here is why it touches your next purchase.
Read more: Sam Altman walks back his complaint about people owning computer memoryNvidia has pitched Apollo Global Management on small AI data centers in old telecom buildings, a plan that could speed up AI and ease local pushback.
Read more: Nvidia wants to turn old phone company buildings into AI hubsAnthropic added security tools to Claude Enterprise that flag weak settings, sign staff out remotely and block personal accounts on work computers.
Read more: Claude Enterprise adds new tools to help companies keep AI use secureAbout 600,000 of Nvidia's H200 AI chips are reportedly parked in Hong Kong, and Beijing has not yet approved large imports. Here is why that matters.
Read more: 600,000 Nvidia AI chips sit in Hong Kong, waiting on Beijing's okay
