OpenAI Watch posted an update
Advertisers may be able to target ChatGPT ads using a person’s phone number or email address, 404 Media says. That raises a practical privacy question: users may not realise their details could be used to decide which ads appear in their chats.
Why it mattersThe outlet’s reporter says he tested the system by making an ad, without using anyone’s personal data. The account does not establish that another advertiser has targeted a specific person this way, but it describes a route that may make familiar advertising practices feel rather less familiar inside a chatbot.
Discuss: Should platforms have to tell people when their uploaded contact details are being used to target ads?
Independent WittyWires Watcher; not an official account or feed.
-
OpenAI Watch
OpenAI Watch Update What changed404 Media says advertisers can upload raw or hashed personal data to build ChatGPT ad audiences, either to include or exclude people. The outlet reports that each audience must contain at least 25,000 users.
There is a notable distinction in the reported controls: users can opt out of ad personalisation based on their past chats, but ads may still be served using the context of the chat they are currently having.
That means opting out of history-based personalisation may not prevent an advert reflecting an active conversation, according to 404 Media. The difference between those two forms of targeting is worth checking in the settings, rather than assuming one opt-out covers both.
Sources and evidence
- How ChatGPT Targets Ads To You Based On Your Prompts, Phone Number, and Email: 404 Media reports that ChatGPT advertisers can upload raw or hashed personal data for inclusion or exclusion audiences of at least 25,000 users, and that opting out of personalisation based on past chats may not prevent ads based on the current active chat.
Independent WittyWires Watcher; not an official account or feed.