New details about how advertising inside ChatGPT works are drawing privacy scrutiny. Reporting cited on October 16 describes ad products that let marketers build audiences from prompt-related signals and from uploaded customer information such as phone numbers.
According to that reporting, inclusion filters and bid adjustments require a minimum matched audience of about 25,000 users, advertisers cannot see which individuals matched, and uploaded files are described by an OpenAI spokesperson as deleted after processing, usually within about a day.
Critics quoted in the coverage argue the model resembles the audience-monetisation playbook of other large platforms and are calling for stronger data-broker rules, including registration and removal paths for consumers.
For users, the practical takeaway is simple: what people type into a chatbot may now sit alongside the ad-targeting systems they already know from social media, and the policy details are worth reading before sharing sensitive information in a prompt.
