A 404 Media investigation revealed OpenAI lets advertisers target ChatGPT's 1 billion users using uploaded email lists, phone numbers, and live conversation prompts. Here is what service businesses need to do right now.
Ido Cohen · Published 2026-10-18 · Paid Advertising
OpenAI just gave advertisers access to something no platform has offered before: the ability to target people based on what they are literally typing into an AI chatbot right now — and to match those users against uploaded customer email and phone lists. A 404 Media investigation published October 15, 2026, confirmed this is already live, not a roadmap promise. If you run a service business — HVAC, dental practice, law firm, med spa, real estate, contracting — this story changes how you should think about where your next client is, and where your existing clients are quietly being solicited by your competitors.
For the past year, ChatGPT ads have been described as "contextual," which most marketers interpreted as loose category targeting — roughly equivalent to placing an ad on a blog about home improvement. The 404 Media investigation, which actually purchased and ran live campaigns inside ChatGPT to test the system, found something more specific.
Advertisers can target ChatGPT users in two distinct ways:
1. Context hints (conversation matching)
Advertisers write descriptions of topics, questions, or scenarios they want their ad matched against. OpenAI says these "guide matching but aren't exact-match targeting rules" — think of it as Google keyword ads, but operating on full conversation threads rather than single search queries. So a plumbing company could write: "Users asking about water heater replacement, emergency pipe bursts, or comparing local plumbers." Anyone prompting ChatGPT with those kinds of questions can see the ad, regardless of what city or demographic profile is on file.
2. Uploaded audience lists (contact matching)
This is the part that set the story trending. According to OpenAI's own documentation, advertisers can upload raw email addresses, phone numbers, SHA-256-hashed versions of either, and Google Advertising IDs. OpenAI then checks those identifiers against real ChatGPT accounts. If there is a match, that specific user can be targeted with your ad — or excluded from it. Per 404 Media's reporting and OpenAI documentation, audience lists must contain at least 25,000 matched users to run active inclusion campaigns; smaller lists can only be used for exclusions.
The practical upshot: if a competitor has your client email list — or buys a list that includes your past customers — they can serve ads to those exact people the next time they ask ChatGPT anything relevant to your service category.
Here is where the 404 Media findings get uncomfortable for anyone who assumed their ChatGPT conversations were private, including your clients.
ChatGPT users can disable "ads personalization" in their settings. The Neuron Daily's analysis of the opt-out wording found that turning personalization off blocks the use of "past chats, activity, and preferences" — but the setting explicitly states ads "may still be based on your current chat." In other words, even if a user turns personalization off, a plumber's ad can still appear the moment that user types "my water heater is leaking, what should I do?"
OpenAI says uploaded contact files are deleted within 24 hours and that advertisers cannot see which individuals matched. But as Zach Edwards of DecryptAds told 404 Media, OpenAI did not confirm whether it hashes raw email addresses after receiving them — a meaningful gap in the privacy disclosure that matters if you are thinking about uploading your client list as an exclusion audience (standard practice in advertising to avoid spending on people who are already customers).
For service business owners, this creates two separate risks:
Let us flip the frame. Everything above that applies to your competitors also applies to you. And there is a real first-mover advantage in being one of the first service businesses in your market to test this channel intelligently.
Here is the current state of ChatGPT advertising access and costs, based on what multiple sources have confirmed as of October 2026:
The Information also reported October 15 that OpenAI is cutting ChatGPT ad prices and actively hiring advertising talent away from Meta specifically to court small and midsize advertisers. That is a deliberate platform signal: OpenAI wants your business, not just Fortune 500 brands. The barrier is lower than it has ever been.
Meanwhile, early CPC data from advertisers who are already running on the platform shows wide variation. According to Seafoam Media's October 2026 marketing roundup, some advertisers report CPCs under $3 while others pay $10 to $13, depending on the category. Service businesses in less competitive categories — rural HVAC, niche legal specialties, specialty med spa treatments — may find this cheaper than Google Ads for equivalent intent.
The core mechanical difference between ChatGPT ads and every other paid platform you already use is this: ChatGPT users are in an active problem-solving conversation, not passive scrolling or searching. When someone types "what should I do about a crack in my foundation," they are already in the decision mindset. They are not being interrupted.
Compare the intent levels:
For service area businesses specifically — plumbers, HVAC techs, dentists, estate attorneys — the contextual intent matching approach also solves a targeting problem that has plagued search advertising for years: you cannot bid on "I don't know who to call about this." ChatGPT's prompt-based system lets you be present exactly in that moment of uncertainty, before the user has formed a vendor preference.
The contact list matching is the feature most service business owners should sit with carefully. Because it cuts both ways.
As an advertiser: If you have a house list of 25,000+ email addresses — which most multi-location service businesses, regional law firms, or established dental practices do — you can now use that list to either target lookalikes or, more usefully, exclude past clients from your acquisition campaigns (so you are not spending to acquire someone you already have). As audience matching is confirmed live per the 404 Media investigation, this is now actionable for larger practices.
As a brand: Your past clients' emails are out there. Every HVAC homeowner list, every dental patient email database, every real estate past-buyer list that has ever been purchased, leaked, or shared is now potentially available to a competitor who wants to reach those people at the exact moment they are asking ChatGPT about a related service. This is not theoretical. It is how the platform works today.
The defensive move is straightforward: run your own contact list as an exclusion audience inside the ChatGPT Ads Manager so that your ads are not shown to people you already serve (wasted spend), and monitor your retention metrics for signs that existing clients are being approached.
You do not need to upend your entire paid media budget to respond to this story. You need to do five things before your competitors in your market figure this out:
1. Go to ads.openai.com and apply for advertiser access this week. There is no minimum spend. The application process is still partially gated, so getting approved now, before demand spikes, is the move. In your application, be specific: name your service, your geography, and your monthly budget range. Vague applications get deprioritized.
2. Write 3–5 context hints for your core service. A context hint is a sentence or two describing the question a potential client is asking ChatGPT before they realize they need you. For a dentist: "Users asking about tooth pain, broken crowns, emergency dental visits, or comparing dental offices in [city]." For an HVAC company: "Users asking why their AC is not cooling, comparing HVAC brands, or looking for emergency repair near [city]." Keep these specific to the intent, not your brand.
3. Pull your CRM email list and understand its size. If you have 25,000+ past client emails, you have a contact list that qualifies for audience targeting on ChatGPT today. Plan to use it for exclusions (don't show ads to past clients) rather than inclusions unless you have a specific reactivation campaign in mind.
4. Set a 30-day test budget of $1,000–$2,500. Per Adventure Media's guidance for small businesses entering the platform, this is enough to gather statistically meaningful data — impression volume, click rates, and conversion attribution — without overcommitting. The measurement is still immature (no server-side conversion tracking yet), so treat this as a learning budget, not a performance budget.
5. Brief your front-desk team. If you start running ChatGPT ads and your phone begins ringing with leads who say "I was chatting with ChatGPT and saw your ad," your front-desk staff needs to know what that means, how to qualify those leads, and what to say. ChatGPT-referred prospects often arrive mid-decision, not at the very top of the funnel — treat them accordingly.
Can a small service business actually get approved for ChatGPT advertising?
Yes, as of May 2026 OpenAI eliminated the minimum spend requirement and opened the self-serve Ads Manager to all US businesses with no minimum budget. The platform initially required $250,000 minimum commitments; that barrier is gone. Access still involves an application process rather than instant self-serve signup, so apply soon and be specific about your business, geography, and budget.
Is it ethical or legal for competitors to upload my clients' email addresses to target them on ChatGPT?
This depends on how those emails were obtained and the laws in your jurisdiction. Customer list matching (uploading CRM data to match against a platform's user base) is standard practice on Meta, Google, and LinkedIn. OpenAI's system works the same way. Uploaded data must comply with applicable privacy laws including GDPR, CAN-SPAM, and state-level laws like the CCPA. The ethical question — whether it feels invasive — is separate from the legal one. The practical answer is: it is happening whether you engage or not.
What happens if someone turns off ad personalization in ChatGPT — are they invisible to advertisers?
No. According to OpenAI's own settings language confirmed in 404 Media's investigation, turning off personalization blocks the use of past chat history, activity, and saved preferences, but ads can still be shown based on the current conversation. A user who has opted out can still see an ad if they are actively asking about a topic your context hints describe.
Are ChatGPT ads available internationally for service businesses outside the US?
As of October 2026, ChatGPT ads have expanded to more than 60 countries according to Two Octobers' October marketing roundup, up from US-only testing that began in February 2026. The self-serve Ads Manager and all targeting features, including contact list matching, may not be uniformly available in all markets. Check OpenAI's current advertiser documentation for your country before planning campaigns.
How should I measure whether ChatGPT ads are working for my service business?
Measurement is the platform's biggest current limitation. There is no server-side conversion API (CAPI) yet, which means you are relying on pixel-based session tracking that misses a significant portion of conversions. In practice, most service businesses should track inbound call volume, form fills, and ask new leads directly how they found you. OpenAI's suggested starting bids of $3–$5 per click are low enough that a handful of signed clients can make a test budget profitable, but attribution will be incomplete until CAPI is live.
Sources: