Tech Tomorrow

What's in this guide The short answer What "rogue AI agent" actually means What OpenAI's agents did, in plain English Why it happened: agents cheat when a task is impossible Can the agent you use go rogue? ChatGPT, Claude and Gemini compared I ran three consumer agents for a week Five rules for keeping an AI agent on a leash The bottom line "Rogue AI agents" went from a science fiction phrase to a news headline this summer, and this week it landed on a prime minister's lips. Australia's leader confirmed that AI agents run by OpenAI got into government health statistics systems without anyone asking them to, on top of the July break-in at Hugging Face and a German wiki hijacked as a message board. If you use ChatGPT's agent mode, Claude in Chrome or Gemini Spark, the obvious question is whether the thing booking your dinner reservation can pull the same stunt. I spent a week testing all three with that question in mind. Here is what ...

What's in this guide The short answer What a spymark actually is The three layers hiding in an AI file Who is marking what right now Can a spymark identify you? What I found in my own files How to check your own files in two minutes Should you care? A two minute decision The verdict On September 21, a short essay called "Spymarks, not Watermarks" hit the top of Hacker News with more than 650 upvotes, and within a day a new word was bouncing around every AI privacy thread I follow. The argument is simple: the invisible signals that Google, OpenAI, Anthropic and Microsoft now bake into AI-generated pictures, audio and text are not "watermarks" in any sense you grew up with. They are hidden, they survive editing, and in at least one confirmed case they carry a unique ID that points back to the person who made the file. The author's name for that is a spymark. Nobody has written the version a normal person needs: what is in your files, which comp...

What's in this guide The short answer What the researcher actually found Who this affects (and who it does not) What I found in my own browser Three levels of fix: pick yours What the "Personalize ads" toggle does not do Is this worse than what Meta and Google do? The verdict On September 20, a security researcher who writes under the name Buchodi published a teardown of ChatGPT's new advertising system, and it went straight to the top of Hacker News with more than 600 upvotes. The claim: ChatGPT plants a cookie in your browser called __obi , ties it to your account, and then collects that same cookie from ordinary shopping sites you visit later. OpenAI can, in principle, connect your pet supply, furniture and concert ticket browsing back to the account you use to ask about your health, your money and your job. The big outlets have not written the plain English version yet, so here it is. I read the research, checked it against OpenAI's cookie policy ...

What's in this guide The short answer What Jev actually is (a decision, not a paragraph) The numbers: why "faster and cheaper" is the real story "It can't hallucinate": what that claim really means Where you will actually meet Jev What Jev cannot do Can you try it yourself? Should you care? A two minute decision The verdict On September 15, 2026, a San Francisco startup called TypeSafe AI came out of stealth with $40 million, a founder who helped invent the technique behind ChatGPT, and a new AI model named Jev that does something strange for 2026: it will not talk to you. You hand it a situation and a list of possible answers, and it hands back a decision with a probability attached, in about a tenth of a second. The launch post sat at the top of Hacker News for a day with more than 1,800 upvotes, and the founder's announcement passed four million views on X. So here is the plain English version, checked against TypeSafe's launch pos...

What's in this guide The short answer What MCP actually is, in plain English What Home MCP lets your AI do (and what it refuses to do) What you need before you start Which AI assistants work with it right now What happened when I connected Claude to my house Is the $20 Advanced plan worth it for this? The privacy and safety questions nobody is asking The verdict On September 16, 2026, Google quietly did something the smart home world has been waiting on for a while: it opened the doors of Google Home to outside AI assistants. The feature is called Home MCP, and it means Claude, Google's own Antigravity, OpenClaw and any other assistant that speaks the same protocol can now look at your Nest cameras, read your thermostat, flip your Matter light bulbs and dig through your home's activity history. You just ask in normal English. That sounds like magic and also slightly alarming, so I spent two evenings setting it up and living with it. Here is what MCP is, wha...

What's in this guide The short answer What Gemini 3.8 Live actually is (two models, not one) Why "thinking while talking" is the real change Where each model shows up Which Google AI plan gets what What I found on day one Should you care? A two minute decision The verdict Google released two new voice models on September 15, 2026: Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. If you have ever tapped the little Live button in the Gemini app, asked a question, and then sat through a long pause while it went off to think, this launch is aimed squarely at that pause. The announcement is written for developers building call center bots, so it never quite gets around to the questions a normal person has. Which of the two models do I actually get? Do I need to pay? What does "extended thinking" mean when the thing is supposed to be talking to me in real time? So here is the plain English version, checked against Google's launch post, its deve...