ChatGPT ads are getting bigger. They still don't buy the answer.

ChatGPT now has two ways a business can show up in the same conversation. One you can buy. One you cannot.

On October 5, 2026, OpenAI published Building advertising for the way people use AI. It announces a new visual ad format, a bigger measurement stack, and brand-suitability pilots. Buried in the middle is the sentence that matters most for anyone who cares about AI visibility:

"As with our other ad formats, advertising does not influence the answers ChatGPT provides."

That is OpenAI's published position. Take it at its word and the operating conclusion is simple: ads buy a labeled slot next to the conversation. They do not buy the answer. The answer is still built from what ChatGPT can find, read, and trust.

Sources opened this fire (all HTTP 200): Building advertising for the way people use AI (OpenAI, October 5, 2026); More ways to measure ChatGPT Ads (OpenAI Ads blog, linked from that post); OpenAI Ads Policies (Updated September 10, 2026; changelog v1.0 March 2026 through v1.6 September 2026); Advertise in ChatGPT (OpenAI Ads landing page).

The short version

• New format. A visual ad format that shows product inspiration, usage, or experiences as images. OpenAI says it will first test it during image generation in ChatGPT, with ads "clearly labeled" and "separate from the image being created." Testing starts later this month in the US with an initial group of advertisers.
• More measurement. Conversion data integrations (Hightouch, Tealium, LiveRamp), attribution partners (AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge, Tenjin), and early incrementality work (Haus, Measured, WorkMagic). OpenAI calls its incrementality work "still in its early stages."
• Brand safety. Negative Phrases for qualifying advertisers, plus suitability pilots with DoubleVerify and Integral Ad Science that test safeguards "without accessing private user conversations."
• Answers stay independent, per OpenAI. The post closes on three principles: "ChatGPT's answers stay independent, conversations stay private, and users remain in control."
• Operator move. Treat ChatGPT Ads as a paid media channel with its own test and its own math. Treat being cited in the answer as a separate, earned channel built on first-party pages. Do not let a vendor sell you one as the other.

What OpenAI actually said

Scale. OpenAI opens with its own reach number: "ChatGPT reaches 1.2 billion people each week." That is OpenAI's figure, not an independent audit.

The format. "Through images showing product inspiration, product usage, or the experiences they make possible, people can discover a new brand, understand what makes an offering distinctive, and decide what to buy." The first placement is image generation, not the text answer.

The separation line. "Ads will be clearly labeled, and remain separate from the image being created. As with our other ad formats, advertising does not influence the answers ChatGPT provides." The Ads landing page repeats it in plainer words: ads are "clearly identified," and "Ads remain distinct from ChatGPT's responses."

The measurement results OpenAI chose to publish. These are partner-reported advertiser outcomes, quoted as OpenAI printed them:

• DV Rockerbox: WeightWatchers' attributed cost per acquisition on ChatGPT Ads was 15.3% lower than its blended paid-search benchmark.
• WorkMagic: statistically significant lift for wellness brand Dose, with 67% of incremental purchases from net-new customers, and an estimated 2.3x more incremental orders than last-click attribution captured.
• Triple Whale: 93% of Portland Leather's visitors from ChatGPT Ads were new.
• OpenAI's own early analysis: 52.7% of eligible one-day view-through conversions happened within an hour of the matched ad impression.

Read those as what they are: a handful of advertisers OpenAI picked, measured by partners OpenAI works with. They are not a benchmark for your business, and none of them is about organic citations.

Who can even buy these ads

This is where local service businesses should slow down. The OpenAI Ads Policies page (Updated September 10, 2026) says that during the initial test period, ads are "primarily limited to consumer verticals such as lifestyle and household goods, local services, travel and experiences, and digital products or education."

Regulated categories are narrower:

• Healthcare and medicine. "Ads for health services outside the US are generally prohibited." In the US, approved advertisers may be allowed "on a case-by-case basis" for listed categories, which include minimally invasive cosmetic procedures. Advertisers "may be required to provide proof of licensure."
• Financial services. Restricted; in the US, case-by-case for approved advertisers, and "Ads for financial services outside the US are generally prohibited."
• Legal services. Permitted "in the US only when the advertiser is licensed to practice law in the jurisdiction where the ad is shown." Outside the US, "currently prohibited." (Changelog v1.5, August 2026.)

The same page also keeps ads away from sensitive conversations, including mental and personal health, and reserves OpenAI's right to decline ads "for any reason," including "business interests, or competitive position" (v1.6, September 2026).

So a dental practice, a law firm, or a financial advisor may not be able to buy their way into ChatGPT at all, or only after manual review. Their path into the answer is still the earned one.

Three things this is not (do not mash them)

1. Not a citation product. Nothing in the October 5 post says ads change which sources ChatGPT cites, links, or recommends. OpenAI says the opposite.
2. Not a text-answer placement launch. The new visual format starts in image generation. OpenAI did not publish that this format will appear inside text answers, in ChatGPT Search results, or in agent modes like Dots.
3. Not proof that organic is dead. No OpenAI page opened this fire gives a share of conversations with ads, a click-through rate for cited sources, or a decline in organic referrals. Anyone quoting those numbers off this announcement is inventing them.

What this means for a New Reward operator

1. Split the two channels on paper. Paid: a labeled ad slot you bid for, measured with pixels, conversion APIs, and incrementality tests. Earned: being the business ChatGPT names and links because your pages answer the question with facts it can verify. Different budgets, different proof.
2. Fix the earned side first if you are in a regulated vertical. If the policy page puts you in a case-by-case or prohibited bucket, ads are not a near-term lever. Your first-party pages, reviews, profiles, and citations are. Start with why your business isn't showing up in ChatGPT.
3. If you test ads, test them like ads. Use a holdout or geo-lift design where you can. OpenAI itself says click-only reporting misses outcomes and that its incrementality work is early. One partner case study is not your forecast.
4. Make the landing page do double duty. OpenAI's policy says ad standards apply to "landing pages" too, and creatives "may not link to destinations that introduce disallowed content." The same page that has to pass ad review is the page an assistant reads when deciding whether to cite you. Clear services, location, licensing, and proof help both.
5. Keep monitoring separate from execution. Watching where you appear, paid or earned, is not the same as shipping the pages and fixes that make you citable. That is the work New Reward does, with before-and-after evidence you can inspect. See GEO has a measurement layer now and GEO measurement is noisy: build a proof loop.

Cloud agents cite who they can see. Ads can put you next to the conversation. Only your own pages put you inside the answer.

Honesty block

• OpenAI did not publish how ads are selected, priced, or ranked, or what share of conversations show ads.
• OpenAI did not publish any effect of ads on citations, links, or recommendations in answers; it states advertising does not influence answers. We cannot independently verify that, and we do not claim to.
• OpenAI did not say the visual format will run outside image generation, outside the US, or inside ChatGPT Search, Dots, or agent modes.
• The 15.3%, 67%, 2.3x, 93%, and 52.7% figures are OpenAI- or partner-reported results for specific advertisers or early campaigns. They are not industry averages and do not predict results for any New Reward customer.
• The 1.2 billion weekly reach figure is OpenAI's own number.
• Category eligibility comes from the Ads Policies page as updated September 10, 2026; OpenAI says these categories "may expand over time" and the list "may evolve."
• Google, Microsoft/Bing, and Perplexity did not confirm anything about ChatGPT Ads in pages opened for this piece.
• No invented caps on how often New Reward customers appear in answers.

Related New Reward pages (live, 200 this fire)

• Why your business isn't showing up in ChatGPT
• OpenAI Dots: proactive research is not citations
• ChatGPT can now use tools on your page
• Reddit didn't vanish from ChatGPT citations
• ChatGPT, Yelp reviews, and local visibility
• AI discovery: sourced is not cited
• GEO has a measurement layer now
• GEO measurement is noisy: build a proof loop
• Become the brand AI knows to cite
• SEO and GEO move together
• Google calls manual fact-check of AI content critical
• AI Visibility Score
• Run a free AI visibility scan
• SEO / AEO / GEO offering · Paid advertising offering · Scan · FAQ