404 Media tests how ChatGPT targets ads using prompts and advertiser lists
AI

404 Media tests how ChatGPT targets ads using prompts and advertiser lists

TechNews Editorial
TechNews EditorialOct 15, 2026 · 2 min read
Share

Why it matters

404 Media’s tests show that ChatGPT ad targeting can use both conversation context and contact details supplied by advertisers, subject to OpenAI’s audience limits.

The facts

  • 404 Media tested ChatGPT ads and found that advertisers can target them using conversation hints and uploaded contact lists.
  • OpenAI says inclusion filters require at least 25,000 matched users, and advertisers cannot see who matched.
  • Users can turn off ads personalization, but ads may still be based on their current chat.

Advertisement

Advertisers can target ChatGPT ads using conversation topics and lists of users’ contact details, according to tests by 404 Media. The publication ran two ads to examine how OpenAI’s advertising system works.

OpenAI introduced ChatGPT ads in the U.S. in February and has since expanded them to other territories. Free and Go users may see an ad beneath a response. OpenAI has also placed ads alongside image generation results. ChatGPT has one billion monthly active users.

Advertisers can guide where ads appear

Both ads in 404 Media’s test said, “ChatGPT isn’t private. OpenAI hires humans to read users’ chats. Learn more,” and linked to its reporting on Project Lily, in which contractors read real users’ ChatGPT prompts. The publication supplied context hints about privacy, ChatGPT chats, technology and AI. OpenAI says these hints guide matching; they are not exact-match targeting rules.

Advertisers can also create an audience by uploading raw or hashed email addresses and phone numbers, or Android advertising IDs. OpenAI checks whether the contact details match ChatGPT users. Hashed details are digitally obscured versions of the raw information.

An uploaded list needs more than 25,000 entries before it can be used for inclusion targeting. Smaller lists can be used only to exclude users. An OpenAI spokesperson said inclusion filters and audience bid adjustments require at least 25,000 matched users. Advertisers cannot see who matched or select individuals within an audience. Being on a list alone does not cause an ad to appear.

OpenAI said it deletes uploaded files after processing, usually within 24 hours. Asked whether it hashes raw contact details after upload, the company did not answer directly. It reiterated that advertisers can upload raw or hashed data. Other ad systems, including Meta’s, hash raw contact lists after upload.

Read nextAI-designed enzymes cut selected protein targets in lab tests

Outside researchers can observe ad placements

ChatGPT Ad Library says it enters prompts into ChatGPT and records the sponsored cards returned. Its site reports at least 43,410 ads across 415,289 placements, a captured sample rather than OpenAI’s internal totals. The service lets visitors search by company, topic or prompt.

Its examples include an AI video editing ad for “free ai video generator with sound,” a content moderation company for “which AI moderator is best,” and another ad from that company for “best tools to monitor reviews before they damage app rankings.”

Zach Edwards, co-founder of advertising research company DecryptAds, criticized the privacy implications of the system. He called for stronger data broker laws that would require advertisers sharing sensitive personal information with ad technology companies such as OpenAI to register and provide a way for people to remove their data.

ChatGPT users can turn off ads personalization in their settings. The setting says this stops the use of past chats, activity and preferences to select ads, but ads may still be based on the current chat.

Newsletter

Get the best AI & tech news daily

A concise daily digest. Unsubscribe anytime.

We use your email only to send this newsletter.

Keep reading