ChatGPT can find who's selling your data

Happy Sunday,

Today I'm showing you how to use ChatGPT to find websites holding your personal information and get your details removed, walking through AI permission settings, and comparing four new AI agents (Dots, Muse, Grok Bot and Instinct).

Let's get into it.

🧠 TRY THIS

ChatGPT Can Now Find Who's Selling Your Data

There are websites you've probably never heard of with profiles containing your address, phone number and other personal details. Some make money selling access to those reports. And you don't have to go through every site trying to figure out its removal process yourself.

You can use the ChatGPT prompts below to have it do the legwork while you work on something else, or do absolutely nothing. Here's the full setup:

Step 1: Open ChatGPT, choose Work at the top, and say:

Answer its questions, then check the matches. Two people can have the same name, so make sure those profiles are actually yours.

Step 2: Confirm your profiles, then stay in the same chat and say:

Step 3: Type @, choose Browser, and give it this:

ChatGPT can then work through the requests with you. Some sites still need an email confirmation or identity check, and removal can take time. The tracker shows what's submitted and what's actually gone.

I also have a scheduled task run at the start of every month to check whether my information has appeared again. You can ask ChatGPT to repeat the check on the first of each month and flag anything that needs attention. Keep new removal requests subject to your approval so you stay in control.

Get my full data-removal guide and prompts →

💬 FROM THE DMS

How to Use ChatGPT and Claude Safely

One of the biggest questions I get is whether it's actually safe to give AI access to your email, files and computer. Many prompts I share need some personal or business information to get useful work done for you.

I completely understand why that feels uncomfortable. You're giving something that can make mistakes the ability to act on your behalf, and that's one of the scariest parts of AI. I do believe these companies take security seriously, but you should understand what you've allowed before handing over more access.

First, check the permission setting. In the ChatGPT desktop app, it's below the message box. These controls help set boundaries around local computer work:

Claude has a similar choice in the message box. The names depend on the version you're using:

Second, pay attention to how it does the work. A direct app connection, often using something called MCP, gives AI specific tools such as finding an email or creating a draft. It doesn't have to guess where to click. Choose an official connection when one is available, but still check the recipient and content because a direct tool can be given the wrong instructions.

Computer use is different: the AI looks at the screen and clicks or types. A misplaced click could mean opening an email or sending it. From my personal experience, I'd only use computer use inside ChatGPT right now. Claude's has been slow and has made mistakes for me. Nothing too major, but I'd keep a closer eye on it.

Third, limit access and save clear rules. Share the project folder it needs and use read-only connections where possible. Training preferences are separate from permission to act. Put this in the persistent instructions for your assistant or project:

Written rules help, but they aren't a lock. Pair them with the actual permission controls. Even Ask for approval won't ask before every edit inside a folder you've already shared.

At the end of the day, AI is a tool you can get a huge amount of value from, and a lot of that value comes from letting it understand your actual life or business. You can share useful context without giving it every account and every permission.

Think about driving a car. You benefit from the safety features, but you still wear a seatbelt and watch the road. Take the same approach here: use the controls, review its answers and pay attention to how it gets things done. We're not at the point where you should stop checking.

Get my full AI safety guide and prompts →

💡 SAVE THIS

Which AI Agent Should You Actually Use? (Dots, Muse, Grok Bot, Instinct)

You've probably heard the terms AI agent or AI employee a lot recently. Over the past couple of months, several major AI companies have launched always-on assistants, and this seems to be their big consumer bet: an AI that keeps working for you after you close the chat.

If you haven't been keeping up, OpenAI released Dots, Meta released Muse, and Elon Musk's SpaceXAI has Grok Bot.

Then there's Instinct, a San Francisco company that announced a $1 billion funding round at a $10 billion valuation on September 28. That came just 33 days after news of its $250 million round at a $2.5 billion valuation. And only a few weeks before that, it had reportedly raised $75 million at roughly a $500 million valuation.

I mean, what is even happening? And yes, those are real reported funding rounds.

It's still in private early access and free to use. That's a lot of money betting on your life admin.

The idea is similar: you give an AI an ongoing job, connect the tools it needs, and it works through tasks instead of just telling you how. They have their own computers in the cloud, so they can keep working when your laptop is closed.

I've tested all four. Here are the pros and cons for each:

After trying them, my personal winner is ChatGPT Dots. It's already connected to the place I do most of my AI work, so my context, tools, messages and voice calls can stay together. If you're already on ChatGPT Pro, that's where I'd start. If you're on Plus, Dots currently needs an eligible Pro plan.

If you're mainly a Claude user or don't want another paid plan, I'd test Instinct once you can get in. I'm working on getting access to share through the newsletter, and if I can arrange it, you'll hear about it here.

For a useful first job, imagine an airline emails you a flight change and it now clashes with your airport transfer. An assistant connected to your email and calendar could flag the conflict and prepare alternatives before you notice. That's the kind of interruption I'd want.

With Dots, create yours in ChatGPT on desktop first, then open the same dot in the mobile app. Message it or tap the phone button to talk. Ask it to review an upcoming trip and flag conflicting times or missing bookings, with your approval before it changes anything.

Set up your ChatGPT dot →

Also, I just got to New York City for the next month! If you're around, I'd love to meet you. Just reply to this email. I'll see it, or my dot will make sure I do lol.

See you Wednesday.

Maverick

P.S. Want your business to actually show up inside AI? The brands that rank are seeing up to 3x the conversions. Get your free AI visibility score and see exactly where you stand inside AI.

P.P.S. If you're new here or you ever skip one, I've got a newsletter archive of all my past emails you can catch up on anytime.