ChatGPT ad targeting reportedly uses uploaded identifiers and chat context
A 404 Media test reports that advertisers can target ChatGPT audiences using uploaded email addresses, phone numbers, or Android advertising IDs, while ad matching can also use context related to users’ conversations.

Uploaded identifiers and conversation context
According to 404 Media, OpenAI’s ad tools let advertisers upload raw or hashed email addresses and phone numbers, as well as Android advertising IDs, to create audiences of ChatGPT users. The report says advertisers can also supply “context hints” to guide ad matching around conversation topics.
It says an audience needs more than 25,000 entries for inclusion targeting; smaller uploaded lists can be used for exclusion. An OpenAI spokesperson told 404 Media that advertisers cannot see which individuals matched and that uploaded files are usually deleted within 24 hours of processing. The report also says that opting out of ad personalization does not stop ads based on a user’s current chat.
The available report does not provide a comparable earlier state for these advertising controls.
404 Media’s tests describe ChatGPT advertising as allowing audience construction from uploaded personal identifiers and contextual matching to chat topics, subject to the reported audience-size limits.
Audience thresholds and limits on individual disclosure
The reported setup links advertising audiences built from personal identifiers with a chatbot environment where ads may also be selected using current conversation context. The report’s stated limits and opt-out language are central to understanding the scope of that claim.
Sources
- 404 MediaHow ChatGPT Targets Ads To You Based On Your Prompts, Phone Number, and EmailRead source →