AI news in context

Agentic bots now account for the majority of internet activity, upending the decades-old assumption that people are the web’s primary users.

• Agentic AI traffic has surpassed human-generated web traffic for the first time.
• The milestone marks a structural shift in internet usage patterns.

Who could have foreseen this?

• Class of AI Models Hyped as Scarily Powerful Apparently Scared the Government Too Much and Now They’re Disabled
• Source-backed details are still developing.

A first-day test of the rebuilt Mac assistant suggests Apple may finally be winning back users who abandoned Siri years ago.

• The Verge published early impressions of Siri AI running on macOS 27 Golden Gate
• The author, a longtime Siri skeptic who avoided Apple Intelligence, said the new version warrants a 'slight' rethink

A Verge writer used Google Gemini to generate a functional backyard app in about five minutes, only to find it paired with an error message declaring the channel 'unrecoverably broken' and a one-click button to repair it.

• A Verge writer generated a functional backyard-organizing app from a single lengthy Gemini prompt in roughly five minutes.
• The working preview appeared alongside an error message stating the channel was 'unrecoverably broken and will be disposed.'

The iPhone’s first native AI editing tools are deliberately conservative, but their reach could redefine mainstream expectations.

• Apple added native AI photo editing tools to iOS 27, including Reframe, Extend, and Clean Up
• Early hands-on tests show the features work as advertised but are conservative compared with Google Pixel AI editing

A new CNET guide explains how to transfer stored AI conversations and preferences into and out of Google’s Gemini.

• CNET published a consumer guide on June 13, 2026 explaining how to move chatbot memory into and out of Google Gemini.
• The guide targets users who want to switch AI services without losing conversation history and personalized context.

A coalition of state attorneys general has formally requested records from the AI lab, intensifying scrutiny over its plan to shed its nonprofit structure.

• A coalition of state attorneys general has formally demanded documents from OpenAI, signaling a multi-state investigation into its corporate activities.
• OpenAI’s plan to fully convert from a nonprofit to a for-profit entity has drawn sustained criticism from California philanthropic, labor, and civic coalitions.

The company is pushing back against a foreign-access block that it says was triggered by a narrow jailbreak finding, arguing the recall of a widely deployed commercial model is an overreach.

• The Trump administration blocked foreign access to Anthropic's Fable 5 and Mythos 5 models over a reported jailbreak vulnerability.
• Anthropic disabled the models for all users worldwide to comply, not just non-U.S. accounts.

Engineers say they were conscripted into a months-old division to write coding puzzles for AI training under an ultimatum and keystroke surveillance.

• Meta's months-old Applied AI division, with ~6,500 staff, is described by engineers as a 'soul-crushing' and 'literal gulag.'
• Employees report being forced into the unit under a 'join or quit' ultimatum to perform monotonous coding-puzzle work for AI training.