Prompt InsightsSubscribe free

Industry2 min read

ChatGPT can show you ads targeted by your prompts, email and phone number

OpenAI lets advertisers target ChatGPT users using their email address, phone number, a unique phone identifier, and the content of their prompts. Most of ChatGPT's one billion monthly active users are unlikely to know this is happening.

A person typing a private message into a chat window while their email address and phone number float visibly in the background.

OpenAI lets advertisers target you inside ChatGPT using your email address, phone number, a unique identifier on your phone, and the content of your prompts, according to 404 Media's own tests of the platform's advertising system. Most of ChatGPT's one billion monthly active users are unlikely to know this is possible.

What happened

404 Media tested ChatGPT's advertising system and found that any advertiser can inject ads into a user's ChatGPT session. To target a specific person or group, an advertiser uploads personal data, such as an email address, a phone number or a unique phone identifier, directly to OpenAI. The ad is then shown to matching users.

On top of that, the targeting can also use the content of the prompts users type. 404 Media has previously reported that human reviewers at OpenAI can read user prompts. This new finding shows those prompts can also feed into ad targeting.

404 Media notes that this kind of targeting is standard practice across the advertising industry. What makes it unusual here is the context: many people treat a chatbot conversation as something closer to a private note than a public action.

What it means for your work

  • If you signed up to ChatGPT with a work email or phone number, an advertiser who holds that data could already be targeting you by name.
  • The questions and tasks you type into ChatGPT, including details about your job, clients or projects, can be used to decide which ads you see.
  • This applies to the free tier of ChatGPT. The 404 Media report does not say whether paid plans are treated differently.
  • The full details of OpenAI's ad targeting options are behind a paywall at 404 Media, so the complete picture of what advertisers can and cannot do is not yet public.

What to do now

  1. Review what personal information is linked to your ChatGPT account, including your email address and phone number, in your account settings.
  2. Avoid typing sensitive details, client names or confidential project information into ChatGPT's free tier if you are not comfortable with that content being used beyond your conversation.
  3. Check OpenAI's privacy policy for any opt-out options related to ad targeting or data use.
  4. If your organisation has a paid or enterprise ChatGPT plan, ask your IT or procurement contact whether that plan carries different data-use terms.

Your ChatGPT prompts are not as private as they may feel, and now they can be used to decide which ads you see.