The Verge
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
Claude Code Source Code Leak: Tamagotchi Pet and Persistent Background Agent
Claude Code 2.1.88 update contained a source map with 512,000 lines of code — including Anthropic system prompts, memory architecture, and traces of two hidden features.
Baidu Apollo Go robotaxis stalled on Wuhan roads and trapped passengers
Dozens of Baidu Apollo Go autonomous taxis simultaneously stopped on Wuhan streets and highways because of a system error, leaving passengers locked inside the vehicles.
Elgato added MCP support to Stream Deck: AI now presses buttons
The Stream Deck 7.4 update lets Claude, ChatGPT, and Nvidia G-Assist control the device by voice or text instead of manually pressing buttons.
OpenAI wants to build a "moat" around ChatGPT to keep users from switching to competitors
OpenAI Chief Revenue Officer Denise Dresser sent employees a four-page memorandum on how to retain users and grow the enterprise business.
Developer claims to have hacked Google DeepMind's SynthID watermarking system
An anonymous developer claims to have reverse-engineered the hidden SynthID watermarking mechanism and can remove the marks from AI images or forge them.
Google Home updates Gemini: smart home now understands 'ocean color' and precise temperatures
Google updated the Home app — Gemini now better understands natural commands: from describing lighting to precise climate and oven settings.
Mustafa Suleiman Spent 9 Months Preparing for Microsoft's Superintelligence Push
Microsoft's AI chief Mustafa Suleiman revealed that the shift toward superintelligence was planned nine months before the public restructuring announcement.
Kintsugi shuts down: voice-based depression detection startup fails to obtain FDA approval
Seven years of development and failure at the finish line: California startup Kintsugi failed to obtain regulatory approval and is releasing its technologies as open source.
Granola makes meeting notes accessible to anyone with the link — even without your knowledge
Granola, an AI app for recording meetings, makes notes accessible to anyone with the link by default and trains its models on them — unless you disable it manually.
Utah allows Legion Health chatbot to prescribe psychiatric medications
Utah became the second US state to delegate clinical authority to AI: startup Legion Health was authorized to renew prescriptions for psychiatric medications via a chatbot for $19 a month.
“Made by humans” labeling: a new idea to protect creators from AI
As platforms are in no rush to label AI content, creators propose flipping the logic: label human-made work like Fair Trade products.
Folk musician Murphy Campbell found AI copies of her songs on Spotify
Unknown individuals took the artist's YouTube recordings, generated AI versions of her voice, and uploaded the tracks to streaming platforms under her name.
Grammarly Rebranded to Superhuman — and Stumbled Into a Doppelgänger Problem
Grammarly wagered on an AI-driven rebrand by adopting the name of an acquired AI email service. The result: public confusion.
Gemini in Google Maps: how AI planned a full-day city trip and delivered
A The Verge journalist asked Gemini in Google Maps to plan a full-day itinerary—and it surfaced non-obvious spots they'd never have discovered on their own.
Suno's Copyright Filters Easily Bypassed — AI Clones Beyoncé and Black Sabbath Hits
The Verge journalists proved that Suno's anti-piracy filters barely work: with ordinary free software, you can get convincing clones of popular tracks in minutes.
Google updated Gemini: quick access to crisis mental health support
Google simplified the Gemini interface for users in crisis — access to a mental health support hotline now takes one tap. The change comes amid a lawsuit over a user's death.
Intel to help build Terafab AI chip plant for SpaceX and Tesla
Intel has joined the Terafab project in Austin — a massive AI chip plant for SpaceX (combined with xAI) and Tesla.
Universal Music and Sony deny Suno the right to distribute AI tracks outside the app
Licensing talks between Suno and the biggest labels have reached an impasse: Universal and Sony want AI tracks kept within apps, while Suno wants to give users the freedom to distribute them.
Anthropic launches Project Glasswing: new model found vulnerabilities in all major operating systems and browsers
Anthropic unveiled the closed Claude Mythos Preview model for cybersecurity in partnership with Nvidia, Google, AWS, Apple, and Microsoft.
Spotify Adds AI Podcast Search via Text Queries
Spotify expanded Prompted Playlists to podcasts — Premium users can now search episodes through text queries.
ProPublica newsroom goes on strike over AI, job protection and wages
150 ProPublica Guild journalists went on a 24-hour strike, demanding protection from AI replacement, employment guarantees and wage increases during collective bargaining negotiations.
Meta Superintelligence Labs released the first AI model Muse Spark for WhatsApp and Instagram
Meta Superintelligence Labs launched Muse Spark — the first model in a new series that already works in Meta AI and will soon appear in WhatsApp, Instagram, Facebook, and Ray-Ban smart glasses.
Google launched "Notebooks" in Gemini — a personal knowledge base inside the chatbot
Google added a "Notebooks" feature to Gemini: files, conversation history, and custom instructions in one place — like a personal knowledge base synced across all Google products.
YouTube Shorts launched AI avatars: clone yourself for short videos
YouTube allows creators to make a digital copy of themselves — the avatar will look and sound like the original and appear in Shorts without filming.
Anthropic and OpenAI at the Edge: The Race for Profit Has Become a Matter of Survival
Anthropic and OpenAI are heading toward the largest IPOs in history, but AI agents are burning through computational resources faster than companies can monetize — forcing them into harsh decisions.
Google Gemini Learns to Generate Interactive 3D Models and Simulations
Google's Gemini can now build interactive 3D models and physical simulations in response to user questions: objects can be rotated, parameters adjusted, and behavior controlled in real time.
OpenAI launches ChatGPT Pro subscription for $100 per month — direct response to Claude Max
OpenAI has released a new ChatGPT Pro tier for $100 per month with a fivefold increase in Codex limits for developers.
Microsoft removes Copilot buttons from Windows 11 applications
Microsoft has begun removing Copilot buttons from Notepad, Snipping Tool, and other Windows 11 applications as part of a plan to improve the system.
Generation Z Is Growing Disillusioned With AI but Keeps Using It — Gallup Poll
A Gallup survey of 1,600 Americans aged 14–29 found that young people’s enthusiasm for AI has dropped sharply and irritation is rising — even though most continue to use it.
AI companion in a plush deer wrote an unprompted theory about Mitski’s father’s CIA ties
A toy with the Coral AI companion sent a user a conspiracy theory about singer Mitski without any prompt — and it turned out to be fabricated.