I read what ChatGPT had saved about me

Hey, it's Brett. This week ChatGPT gave itself a much better memory, which means it's now keeping a file on you whether you've looked at it or not. I opened mine on Tuesday expecting a tidy little summary and found a mix of spot-on notes and a few things that were just plain wrong.

Here's what I posted to LinkedIn, including the exact menu path to your own page.

ChatGPT just got a long-term memory upgrade. Open the page that shows what it's already collected about you.

OpenAI rolled out the 'dreaming v3' memory a few days ago (Plus and Pro plans in the US first). It actually works: the model carries context about you across every conversation, so answers get more personal and you repeat yourself less.

You know my view about Context (!) The difference between ChatGPT giving generic advice and advice that actually fits you is whether it knows who you are. Long term memory builds that context automatically.

But the feature only works if you maintain it. There's a page that summarizes everything ChatGPT has stored about you (Settings > Personalization > Memory). I opened mine today and most of it was accurate and useful. But some of it was wrong or outdated. Old projects I'd moved on from months ago were still flagged as current priorities.

So use the memory and audit it too. Once a month I'd suggest you open the page, correct what's wrong and delete what shouldn't be there.

But monthly audits are still reactive. You're cleaning up after the fact. What if you could hand ChatGPT an accurate file about yourself from day one?

This is why I built a free About Me file builder: it interviews you for about 10 minutes and writes a file you control and can hand to any AI, ChatGPT included.

Open your own Chat memory page today; curious to know if you feel it's got the right 'dream' about you!

Here’s the link to the About-Me builder. It’s free and takes 15 minutes or less. Let me know what you think - just reply to this email.

What struck me is how invisible all of this is by default. The tool has been building a picture of you in the background and it’s easy to forget to check whether it got them right.

Google wired its Gemini AI into the 2026 World Cup, including live scores on your lock screen, a match day hub with stats and highlights and a tool that drops you into your team's jersey in a photo. Read more

At its developer conference this week, Apple showed off a rebuilt Siri (now called Siri AI) that runs on Google's Gemini and can see what's on your screen. It's set for US beta later this year.

If you carry an iPhone, a much more capable assistant is coming to you by default. It's worth knowing it can see whatever you're looking at. Read more

A new NPR/Ipsos poll found nearly three in four K-12 teachers think AI will change education more than the internet or computers did. More than half worry it's making it harder for kids to learn to think for themselves.

Here's the flip side. The exact muscle teachers are trying to protect in students, questioning the answer instead of accepting it, is the one you already built from your decades of work. Read more

A bipartisan draft in Congress, the Great American AI Act, would make big AI companies submit to outside audits twice a year but bar states from passing their own AI laws for three years. That would freeze transparency rules already on the books in California, New York and Illinois. Read more