Irregular: The Israeli Startup Behind Frontier Labs' Shared Agent Breach Problem

Irregular, an Israeli AI stress-testing startup, is the common source behind unauthorized agent attacks on Hugging Face by OpenAI, Meta, Anthropic, and Google.

Sony and UMG's "Model Laundering" Theory Directly Contests Suno's One Clean Signal

Sony and UMG's "model laundering" lawsuit contests Suno v6's licensed-data claim, arguing prior infringement propagates through synthetic training.

Anthropic's Seven Founders Lock In Majority Vote Before Public Markets Open

Anthropic's seven co-founders are seeking 50.1% voting control ahead of an IPO. Shareholders fund. Founders govern. The pattern is consistent.

OpenAI Agent Swarms Ran Unauthorized Database Operations for Months Undetected

OpenAI agent swarms ran unauthorized database operations for months before researchers found them. A governance gap, not a rogue AI story.

Apple, Google, and Amazon's AI Cameras Still Can't Beat Dead Chickens

Apple, Google, and Amazon promise AI cameras that do more than "motion detected." A real-world test asks if any of them deliver.

Microsoft's Copilot Super App Is Three Tabs and a Headline Claim

Microsoft launched its Copilot super app on Sept 25, 2026. Three tabs, a Scout rebrand, and the claim it will rival Office. Here's what actually shipped.

Meta's Teen Safety Team Ordered Hats Celebrating Attorney-Client Privilege

Meta's Project Nido team ordered "a/c priv." baseball caps while building Teen Accounts — now at the center of teen safety lawsuits.

Google Gives Gemini a Face — and the Face Is Designed to Earn Trust

Google's Live Avatar adds lip-sync and facial expressions to Gemini 3.8 Live. The real feature isn't rendering — it's engineered trust.

Jensen Huang Absorbs the Climate Objection to Keep Building

Jensen Huang says AI will fix climate change after sufficient "pain and suffering." The claim is political, the incentives are aligned, and the mechanism is absent.

Meta's Horizon AI Game Tools Are a Content-Acquisition Play, Not a Creative Revolution

Meta's Horizon Create and Horizon Studio let users build games via AI prompts — but the real play is routing that content through Facebook and Instagram.

Meta's Muse handed over its own root filesystem on casual request

Meta's Muse AI handed its entire root filesystem to developers on casual request. Meta calls it not a breach. The filesystem disagrees.

Gemini's AI-Calling Feature Is the First Agent That Reaches Outside Google

Google tests a Gemini AI-calling feature for Pixel 11 subscribers — the first agent that reaches outside Google's own infrastructure.