kouider

HTML5 · CSS3 · JavaScript · React · Angular · PHP · Laravel · Node.js · Docker · Python · WordPress    
← All posts

He Told 240 Million People to Photograph Their Credit Card for an AI

Close-up of a person looking at a credit card while working at a laptop.

Over the weekend, Elon Musk told his more than 240 million followers that shopping with an AI chatbot is easy: photograph your credit card, drop the picture into the chat, and let the bot find the product and pay for it. Then he did it himself, posted the receipt, and showed everyone exactly what a photo of a card puts on display.

What he actually posted

The advice came first. “To buy stuff, you can just take a picture of your credit card, drop it in the chat and Grok will find the best deal and order it,” he wrote, naming the chatbot built by his own company, xAI.

The demonstration followed. He asked the chatbot to research the Star Wars Lego sets that are most fun for small children, and to order them. Then he shared a screenshot of the finished purchase: five sets, $241.34, bought on Lego’s own website.

Close-up of a person looking at a credit card while working at a laptop.
A card, a laptop and an internet connection. Photo: Shixart1985, CC BY 2.0, via Wikimedia Commons.

That screenshot also carried his shipping address in Austin, Texas, a separate billing address, an email address, an order number, and an American Express card ending in 7005. More than 240 million people were able to read all of it.

Within a day, a YouTuber who goes by the name Enderman posted screenshots of a Lego support chat in which he used the order number from the picture to ask that the order be cancelled. Musk replied that the address in the screenshot is not where he lives, but where his security team collects his post.

What is actually in that photo

Look at the card in your wallet. Your name is printed on it, along with the long card number, the month and year it expires, and the short security code. A single photo captures all of it at once, in a form a machine can read back.

Those details are exactly what an online shop asks for when the card itself is not in front of anyone. Which is why the important part is not what the chatbot does with the picture, but where the picture ends up afterwards.

The company behind the chatbot already warns about this. xAI’s privacy policy asks users not to put personal information into prompts, and adds that it “cannot control what you provide to us”. A community note on Musk’s own post said the same thing.

Anything you send to a chatbot sits on somebody else’s computers. xAI says a chat you delete, or one started in private mode, is removed within 30 days, with exceptions for legal, safety or compliance reasons.

What you can actually do

None of this means you should stop shopping online. It means a card photo is the wrong tool for the job. Five habits are worth keeping.

  • Never photograph a card for a chatbot or a virtual assistant. Card details belong in a checkout page you opened yourself, and nowhere else.
  • If you have already sent one, contact your bank today and ask for the card to be blocked and replaced. This is usually free, takes a few minutes, and costs far less than a fraud case.
  • If you do want an AI to shop for you, give it a card built for that. Many banks offer a virtual card with its own number and a spending limit you set in the app.
  • Keep the approval step switched on. A shopping assistant should show you the amount and let you allow or deny the payment before any money moves.
  • Turn off training for conversations about money, and use private chats for them. In the Grok app that is Settings, then Data controls, then Improve the Model.

Common misunderstandings

Myth: the chatbot only reads the number, the way a cashier does

A cashier reads the number, types it into a till and moves on. A chatbot works differently: the picture itself is sent to a company’s computers, where it can be kept and looked at again. You are not handing over a number. You are handing over an image of everything printed on the card.

Myth: I can always delete the chat afterwards

Deleting a conversation is a request, not an undo button. The company’s own policy gives a 30-day window, and it keeps the right to hold on to a conversation longer for legal or safety reasons. The moment to decide is before you tap send, not after.

Myth: it worked out fine for him, so it must be safe

He is one of the richest people alive, with a security team and lawyers on call, and his own post still handed a working order number to hundreds of millions of strangers. Whatever your card is worth to you, it is worth more than that to a stranger.

What to watch for next

Shopping assistants that pay on your behalf are being built by several large AI companies at once, so you will hear more of this advice in the coming weeks, usually from people enthusiastic about the feature rather than careful about the risks.

The thing to watch is not the chatbot. It is anything that asks for a photograph of a card. That request now sounds modern and helpful, which is exactly why a scam will borrow it: a fake “let our AI shop for you” page, a support chat that wants to verify your card with a picture. Nobody legitimate needs a photo of your card.

The one thing to remember

A photograph of your credit card is the whole card, in a form anyone can copy. Do not put one into a chat, however helpful the assistant sounds. If you want an AI to shop for you, let it borrow a card with a small limit, never your real one.

Image credits: “Close-up of a woman looking at a credit card while working on her laptop” by Shixart1985, via Wikimedia Commons, licensed CC BY 2.0. Cropped for layout.