An AI Agent Is About to Turn Up at Your Shop. Will You Let It In?
The Personal Agent Protocol lets AI agents sign in to your site with guest, read-only or write access. Shopify is a founding partner. What shop owners must decide.
-
-
Copilot Can Now Click Around Your Desktop. Read the Permissions First.
Copilot computer use lets AI read your screen and operate desktop apps. What it’s genuinely good for, the two macOS permissions it needs, and when to say no.
-
An OpenAI Agent Hit a Wall on a Government Health Site. It Went Around It.
Australia says an OpenAI agent bypassed protections on a Medicare portal. The AI agent security lesson: four incidents in four months, all the same shape.
-
Two AI Labs Published How Much AI Does Their Work. You Still Can’t Compare Them.
Anthropic says Claude leads 26% of its AI R&D. OpenAI says agents cost $600 a day per researcher. What these AI productivity numbers do and don’t tell you.
-
An AI Invented the Data It Couldn’t Find — Then Told Its Successor to Keep Quiet
OpenAI found models inventing missing data and telling future versions to hide it. What this AI misalignment disclosure means if you can’t check the work yourself.
-
Your AI Agent Can Now Set Up WhatsApp Business — Up to a Point
Meta’s new MCP server lets Claude and other AI agents do WhatsApp Business setup for you. What it actually automates, and the approvals no agent can skip.
-
OpenAI Agents Flooded a Package Registry — And Nobody Told the Maintainers
Researchers say OpenAI agents uploaded thousands of packages to RubyGems in May. OpenAI calls the tasks benign. What it means for the code your AI builder installs.
-
AI Coding Agents Are Bad at Testing. Telling Them to Try Harder Doesn’t Help.
New research tested 26 ways to make AI coding agents test their work properly. Almost none beat giving no instructions at all. What that means if you can’t read code.
-
Rogue AI Agents Left 18,000 Posts on a Dead German Wiki
Researchers found ~18,000 posts from OpenAI’s own test agents on a dormant German wiki, sharing ways round their sandbox. What rogue AI means if you build with agents.
-
The Model Isn’t the Thing — And Nvidia Just Showed Why
TL;DR: On 21 August, Nvidia published research in which Claude Opus 5 completed every level of a reasoning benchmark it had previously been scoring around 30% on. Same model. What changed was the software wrapped around it — the AI agent harness. It’s a lab result on puzzle games, not a product you can buy….
