
ChatGPT lets advertisers target ads with emails and phone numbers
OpenAI documents that brands upload customer lists to include or exclude people. The ads appear only on the Free and Go plans.
Advertisers on ChatGPT can upload lists of their customers' email addresses and phone numbers to aim a campaign at those people or to leave them out, according to OpenAI's help center. The feature is called custom audiences and is set up in Ads Manager, the company's campaign tool.
The lists accept five types of identifier:
- Email address.
- Phone number.
- Email hashed with SHA-256, a cryptographic fingerprint that hides the original data.
- Phone number hashed with SHA-256.
- Google advertising ID.
To target by inclusion or adjust how much a brand bids, at least 25,000 matched users are required. With fewer, the list can only be used for exclusions, for example to avoid showing an ad to someone who already bought. The file can be up to 500 MB, in CSV or TXT format.
OpenAI says Ads Manager does not show which individual users match and does not let advertisers pick specific people. On its page about ads in ChatGPT it also states that it does not share conversations with advertisers or sell data. Those are the company's claims, and there is no independent verification.
Ads appear only on the Free and Go plans. They are not shown to users under 18 and do not exist on Plus, Pro, Business, Enterprise or Edu. Tests in the United States began on Feb. 9, 2026. On Oct. 15, 404 Media published its own tests showing the platform targets ads using email, phone number or mobile device ID, and noted that ad selection also uses what the user types in the chat. This outlet could not read the full text, which is behind a paywall.
OpenAI offers three controls in Settings, under Ads controls: turn off personalization, delete the data used for advertising, and see why each ad appears. With personalization off, ads continue, now based on the context of the current thread.
It is not yet clear whether custom audiences already work in Mexico. Local outlets reported a pilot for the Free and Go plans in August, but their dates differ. Mexico's Anti-Corruption and Good Governance Ministry took over data-protection duties from the now-defunct INAI; this outlet found no statement from it about these lists.
Keep reading
- Privacy
Cisco patches exploited 9.8-rated Secure Email Gateway flaw
- Artificial intelligence
Eight U.S. states sue OpenAI over ChatGPT and minors
- Privacy
Microsoft takes down YAM Store, a phishing kit market
- Artificial intelligence
FTC sends investigative demands to OpenAI, Anthropic, Axios says



