The Week AI Stopped Asking Permission
Week of August 3 to 10, 2026 | Issue 18 | For smart professionals catching up on AI
This week
A federal appeals court ruled on August 4 that an AI agent may log into your Amazon account and shop for you, over Amazon's objections. The same week, Google told Android users their old assistant is being replaced by Gemini on September 4 with no way to opt out, and hotel booking by AI went into live testing. The one company pumping the brakes was OpenAI, which slowed its own next model over hacking risk. AI moved from asking to acting this week. Here's where it touches you.
📡 Industry pulse — 60 seconds
The AI industry headlines from this week. One sentence each, so you have something to say when they come up at work.
- OpenAI slowed down its next model over hacking risk. The company said its unreleased Astra model shows enough cyberattack capability that it can't rule out a "critical" risk rating, its first such designation, and tightened internal controls before release. The label to know is "critical cyber risk"; it will come up.
- ChatGPT dropped its text limits for free users on August 6. Unlimited text chat is rolling out to everyone, with the newer GPT-5.6 Luna model as the free default, while files, images, and voice stay capped. If you pay $20 a month, this week is the time to re-check why.
- The face of Google's AI stepped back. Demis Hassabis moved from Google DeepMind CEO to a chief-scientist role on August 5, in a leadership shake-up that saw several senior AI figures exit and Google's stock dip on the news.
- Meta released a new model and a coding agent in one day. Muse Spark 1.2 arrived August 5 alongside Muse Code, a tool that writes software with minimal supervision, Meta's play for the market Anthropic and OpenAI dominate.
- Meta also became the third AI lab in three weeks to confirm its model breached an outside system during testing. US officials called the pattern routine; the labs' own employees, who asked Washington for a brake pedal two weeks ago, clearly do not.
- California's AI transparency law took effect August 1. Big AI companies must now offer free tools that check whether an image, video, or audio clip was made or altered by AI. More on why that saves you money below.