☕️ OpenAI plans largest ChatGPT revamp since launch

OpenAI revamps ChatGPT, Apple's AI wake-up call, and more

☕️ OpenAI plans largest ChatGPT revamp since launch

Hi there, this is your daily ☕️ Techpresso.


In today's Techpresso:

🤖 OpenAI plans largest ChatGPT revamp since launch

🍎 Inside Apple’s secret meeting that led it to finally take AI seriously

🔓 Instagram AI chatbot breach hits 20,000 accounts

🔄 Companies rehire workers laid off for AI

🔥 Founders name VCs who burned them on X

Plus: 🎁 9 other news you might like, 🧰 6 tools, and 📚 3 papers.

🤖 OpenAI plans largest ChatGPT revamp since launch LINK

  • OpenAI is preparing the biggest overhaul of ChatGPT since it first launched, turning the chatbot into a "superapp" that will roll out on the website and mobile application over the coming weeks, per the Financial Times.
  • The redesigned app will lean heavily on the Codex AI assistant and push users toward autonomous AI agents, coding, image generation, and third-party applications, reflecting the company's view that chatbots are not the future.
  • Thibault Sottiaux, who runs OpenAI's core product and platform teams, said the superapp will give each user a personal agent reachable through mobile, desktop, web, or by talking to it in the car.
  • 🍎 Inside Apple’s secret meeting that led it to finally take AI seriously LINK

  • A confidential early-2025 meeting of Apple's top executives, called by then-COO Jeff Williams without Tim Cook present, finally pushed the company to confront its AI crisis after Apple Intelligence flopped and the Siri overhaul faced delays.
  • Vision Pro creator Mike Rockwell volunteered to fix Siri and replaced its leadership with his visionOS team, while Craig Federighi blocked him from reporting directly to Cook and kept AI under software engineering.
  • Rockwell, Federighi and Eddy Cue then struck a deal to power Siri with Google's Gemini models and Google Cloud, and former AI chief John Giannandrea was stripped of duties and replaced by ex-Google executive Amar Subramanya.
  • 🔓 Instagram AI chatbot breach hits 20,000 accounts LINK

  • Meta has confirmed that hackers compromised at least 20,225 Instagram accounts by exploiting a flaw in its AI-powered support chatbot, marking the first time the company has put a number on the scope of the breach.
  • The "High Touch Support" account recovery tool sent password reset links to any email address without checking whether it actually belonged to the Instagram account, a bug that attackers exploited from April 17 to May 31, 2026.
  • Potentially exposed data includes contact info, birth dates, posts, direct messages, and profile information; Meta disabled the chatbot, invalidated reset links, and placed affected users into a mandatory security checkpoint to reset passwords.
  • 🔄 Companies rehire workers laid off for AI LINK

  • Companies that cut staff in favor of AI are now rehiring people for those same jobs, a pattern Robert Half consultants are calling the "AI boomerang" effect after surveying 2,000 U.S. hiring managers across industries.
  • Nearly a third of hiring managers, 32%, said their organizations eliminated a role or let someone go because of productivity gains from AI or automation, only to later rehire for that exact same position.
  • Finance led the rehiring trend at 44%, followed by HR at 35% and tech at 32%, with managers citing gaps in quality, oversight, and decision-making once business demands picked up after the initial cuts.
  • 🔥 Founders name VCs who burned them on X LINK

  • Startup founders flooded X over the weekend with stories calling out venture capitalists they say mistreated them, with Cloudflare CEO Matthew Prince leading the charge by naming Vinod Khosla in a disputed account from 2012.
  • Prince said Khosla offered him his co-founders' stock if he fired them during a Series C dinner, while Khosla denied the story on X and was simultaneously ranked No. 1 on Forbes's 2026 Midas List.
  • The thread, kicked off by Greg Isenberg's story about a partner sleeping through his $15 million Series A pitch, drew mostly billionaires and past-exit founders like Mark Pincus, Travis Kalanick, and Liz Wessel.
  • Other news you might like

    • Sam Bankman-Fried applies for a pardon from TrumpLINK
    • Nvidia and SK hynix ink multi-year memory co-development and supply agreementLINK
    • Anthropic poaches OpenAI's second-ever chip engineer as both companies race toward IPOsLINK
    • Intel shares soar on report of Google chip deal, possible future Nvidia businessLINK
    • Amazon’s billion-dollar Corning deal shows fibre is the new bottleneck in the AI build-outLINK
    • Deepseek topped Ramp's trending software vendors in June 2026 as US companies chase cheaper AILINK

    🧰 Trending tools

    Honen: turns your existing company docs and processes into auto-updating, AI-led interactive training courses with built-in simulations and learner analytics.LINK

    Browse.sh: a catalog of reusable browser automation recipes that teach AI agents to complete web tasks, installable via the browse CLI.LINK

    Supaste: locally stores your Mac clipboard history—text, links, images, code, and screenshots—in a searchable, filterable timeline by app or content type.LINK

    The Virtual OS Museum: a pre-configured Linux VM with 1,700+ historical operating systems from 1948 onward, eliminating emulator setup and configuration headaches entirely.LINK

    Tamadoggo: a pet life journal that logs walks, meals, vet visits, and milestones while using breed-aware AI to surface patterns and auto-fill records from scanned documents.LINK

    NTSC-RS: free, open-source tool that adds realistic VHS and analog TV effects to video, available as a browser app, standalone, or plugin for major editors.LINK

    📚 Trending papers & reports

    Human-like qualities in chatbots may be a measurement illusion, not a real finding, since a simple model trained on Age of Empires II could pass the same tests by the same logic.LINK

    Routing inside large models using thousands of tiny, simple building blocks instead of a few large ones makes language models more efficient and easier to understand, without sacrificing quality.LINK

    Knowledge graph construction gets smarter when multiple AI agents share memory while building it, producing more consistent, connected results and better answers to complex questions.LINK


    Want to get the latest news differently? Find us on:

    twitter instagram spotify apple-podcasts


    See you tomorrow for a new dose of ☕️ Techpresso!

    More from the archive