ChatGPT Ads Can Target You by Email, Phone
ChatGPT advertising just got a lot more concrete. [404 Media's October 16 test](https://www.404media.co/chatgpt-ads-target-your-phone-number-email/) bought two real ads through OpenAI's self-serve platform and confirmed advertisers can target ChatGPT users by uploaded email address, phone number, or Android advertising ID, raw or hashed, and by the actual content of what someone types into the chat. The underlying custom audiences tool is not brand new, OpenAI rolled it out for beta advertisers in July 2026, but this is the first independent test confirming exactly how granular the targeting goes, including the prompt-based matching. That is the part worth a marketer's attention.
What 404 Media actually tested
404 Media purchased an ad reading "ChatGPT isn't private. OpenAI hires humans to read users' chats. Learn more," and set context hints, OpenAI's term for topics or keywords an ad should match, targeting people asking ChatGPT about privacy. The ad showed up in exactly that context. Separately, the outlet confirmed advertisers can upload lists of emails, phone numbers, or device identifiers to build or exclude an audience, with OpenAI checking the list against real ChatGPT accounts. OpenAI requires a minimum of 25,000 matched users per audience and says uploaded files are typically deleted within 24 hours.
This feature already existed, the test is what changed
Custom audiences for ChatGPT ads launched for beta advertisers in July 2026, reported at the time by MediaPost and Jon Loomer Digital, alongside suggested ad drafts and expansion into Japan and South Korea. None of that is new as of this week. What is new is public confirmation, via a real purchased ad, that the targeting works the way OpenAI's documentation describes, and that it extends to matching on the content of a user's actual prompts, not just uploaded contact lists.
OpenAI's targeting filters require at least 25,000 matched users per audience and advertisers are not shown which specific people matched. OpenAI also says ads can still draw on a user's current conversation for relevance even with personalization turned off.
What this means for a marketer
If you already run ChatGPT ads or are evaluating the channel, nothing in your buying process changes today, this confirms the mechanics work as documented, it does not announce a new product. The real takeaway is about trust and disclosure. Raptive's October 2026 consumer survey, covered in an earlier Signal post, found that naming a trusted source nearly doubles consumer trust in an AI answer. The same logic cuts the other way for ads inside a chat interface: a user who finds out mid-conversation that their prompt just got matched against an ad campaign is a different trust problem than seeing a banner ad on a search results page, and it is worth watching how OpenAI handles that disclosure as this channel grows.
What to do this week
Nothing urgent to act on in Ads Manager equivalents here, there is no new buying mechanic to adopt. The practical move is the same one that applies to any ad platform accepting a customer list: treat an uploaded audience file with the same data-handling care as your CRM export, confirm your own privacy policy covers third-party ad matching if you plan to use this, and keep an eye on how OpenAI communicates this to end users, since that disclosure question is still open and could shape whether this channel scales the way Meta and Google's ad businesses did.
FREQUENTLY ASKED
Can ChatGPT ads target me by email or phone number?
Yes. OpenAI lets advertisers upload lists of email addresses, phone numbers, or Android advertising IDs, raw or hashed, then checks them against ChatGPT accounts to build a matched audience. A 404 Media test confirmed this works in practice, with a minimum of 25,000 matched users per audience.
Does ChatGPT use what I type to target ads?
Yes. Advertisers set context hints, topics or keywords an ad should match, and OpenAI says those guide ad placement based on the conversation. 404 Media confirmed this by buying an ad targeted at people asking ChatGPT about privacy and having it appear in that exact context.
Is this a new ChatGPT feature?
The underlying custom audiences tool launched in July 2026 for advertisers in the ChatGPT ads beta. What changed on October 16 is that 404 Media independently tested and confirmed exactly how granular the targeting is, including prompt-based matching, which had not been verified publicly before.
What should a marketer do about this now?
Nothing changes in how you buy ChatGPT ads today. The useful move is tracking how OpenAI discloses this to end users going forward, since trust in AI answers is already a measured factor, and treating any customer list you upload to any ad platform, including this one, with the same care as your CRM data.
RELATED SERVICES
Delivered in 22+ markets worldwide.
Want this done for your business?
Free audit. No pitch. 24-hour turnaround.
PROOF THIS WORKS