Claude shows you what it remembers, AI starts clicking for you, and a text to ignore: This Week, Plainly #4

Hello, and welcome to this week’s This Week, Plainly. A lot of this week’s news is about control: what your AI remembers, what it forgets, and how much it’s allowed to do on its own. Here’s what happened, in plain English, and whether you need to do anything.

1. Claude now shows you everything it remembers, topic by topic

What happened: On 25 August Anthropic, the company behind Claude, changed how Claude’s memory works. Everything Claude remembers about you is now listed as short notes, grouped by topic, and you can read, edit or delete each one. Claude also now adds to its memory as you chat, rather than afterwards. Memory is on by default on the Free, Pro and Max plans.

By default, Claude doesn’t save sensitive subjects such as your health, religion or politics. A new setting, include sensitive topics in memory, lets you change that if you’d like it to remember, say, a food allergy.

Should you care? 🟢 Yes, it’s worth a two-minute look. Go to Settings → Memory and open the list under Topics. Fix or delete anything wrong or out of date. You can pause or reset memory from the same place. We’d leave the sensitive topics setting off unless you have a clear reason to turn it on.

2. ChatGPT’s temporary chats get a few more choices

What happened: A “temporary chat” in ChatGPT is one that doesn’t appear in your chat history and doesn’t add anything to ChatGPT’s memory of you. On 27 August OpenAI, the company behind ChatGPT, started rolling out two new options. When you start a temporary chat, you can now choose to let it use your saved memories and custom instructions, so the answers still suit you. And if a temporary chat turns out to be useful, you can save it as a normal chat.

Should you care? 🟠 Only if you already use temporary chats. Nothing changes unless you choose it: temporary chats still start out not personalised. One thing to know: if you save a temporary chat, it becomes an ordinary chat and follows your usual settings, including memory. So if you used a temporary chat for something private, simply don’t save it.

3. AI that clicks, types and fills in forms for you

What happened: On 26 August Anthropic made Claude in Chrome available on every paid Claude plan. It’s an add-on for the Chrome web browser that lets Claude read the page you’re on and click, type and fill in forms, using the accounts you’re already signed in to. The day before, OpenAI said ChatGPT Work (on the Plus and Pro plans) can now sign in to some websites and carry on with a task while you step away.

Should you care? ⚪ Not yet, for most people. Both are paid extras, and both companies say the AI checks with you before anything important, like a payment. But Anthropic itself warns about “prompt injection”: hidden instructions in a web page or email that try to trick the AI into doing something you didn’t ask for. For now, let AI help you understand websites and forms, and keep the clicking, paying and signing in to yourself.

4. Google’s Gemini Live can now read your emails and plan your day, out loud

What happened: Gemini Live is the part of Google’s Gemini app you talk to out loud. On 26 August Google added a spoken “Daily Brief” that sums up your day from Gmail and Google Calendar, hands-free email sorting (“any new emails?”), and memory of earlier chats. It can also hand bigger jobs, like keeping a weekly meal plan going, to a background helper called Spark.

Should you care? ⚪ Only if you already use Gemini with Gmail. Daily Brief needs a paid Google AI Plus plan or higher, and Spark needs Google AI Pro or higher. Google hasn’t said which countries get each feature, so it may not reach your phone straight away.

5. 🔴 Scam watch: “a standing order has been set up on your account”

What happened: Which? is warning about a scam text pretending to be from Nationwide. It says a standing order (a regular payment) has been set up on your account, and gives a number to call if it wasn’t you. That number goes straight to a fraudster after your personal details. Which? has also seen scam calls pretending to be from the mobile network Three, saying your number will stop working within 24 hours or your SIM card will be blocked.

Should you care? 🔴 Yes, this is the one to remember. Both scams work by making you panic about your money or your phone.

Never call a number given in a text or email you weren’t expecting
Check your account in your bank’s own app, or call the number on the back of your card
If a caller says your phone will be cut off, hang up and contact your network yourself
Never read out a security code sent to your phone

Forward scam texts to 7726 (it’s free). Report fraud at reportfraud.police.uk or call 0300 123 2040, and tell your bank straight away if you’ve shared any details.

💡 Tip of the week

Tell it who it’s for. The same question gets a very different answer if you add “Explain it for someone who’s never used a smartphone” or “Write it for a ten-year-old”. Saying who the answer is for is one of the quickest ways to get something you can actually use.

That’s it for this week. Know someone who’s curious about AI but not sure where to start? Forward this on. It’s free, and they can sign up at deadeasyai.com/news.