OpenAI to test visible ads under ChatGPT image generation in the US from October
The announcement came on October 5, 2026: OpenAI is launching a new visual ad format shown during image generation in ChatGPT, with testing in the US later in October for an initial group of advertisers. At the same time, a full adtech stack is being built out — conversion measurement, attribution, incrementality and brand suitability — in the clearest signal yet that advertising is a core business line for a product the company says reaches 1.2 billion weekly users.
What was announced
OpenAI will initially test the format during image generation in ChatGPT, according to BleepingComputer (Mayank Parmar, October 5, 2026). The company says testing begins later in the month in the US with an initial group of advertisers, but neither the advertisers, the exact start date nor the pricing model have been disclosed.
OpenAI gave the following rationale, as reported by BleepingComputer (quote translated from English): "Initially, we'll test this new ad format during image generation in ChatGPT. Ads will be clearly labeled, and remain separate from the image being created."
Why ads — and why now
OpenAI says ChatGPT now reaches 1.2 billion people each week, and that advertising becomes a new way to monetize users who do not pay for subscriptions. AI Weekly (Alexis Dufresne, October 5, 2026) characterizes this editorially as OpenAI's "first real monetization of the free tier through ads," a new revenue line alongside subscriptions. That is an analysis, not a fact the company itself has established.
The point is that OpenAI is pulling an established adtech stack — LiveRamp, DoubleVerify and IAS, according to AI Weekly — onto a platform with a user base in the billions: 1.2 billion weekly users, according to the company itself.
The measurement infrastructure behind it
According to Unite.AI (Aiden Cross, October 5, 2026), the expanded ChatGPT Ads ecosystem includes:
- Conversion data: Hightouch, Tealium and LiveRamp
- Attribution: AppsFlyer, Adjust, Branch, Triple Whale, Kochava, DV Rockerbox, Northbeam, Singular, Airbridge and Tenjin — with support for Conversions API, reporting and click attribution
- Incrementality: Haus, Measured, WorkMagic, Kantar and Cint
- Brand suitability: pilots with DoubleVerify and Integral Ad Science
The size of the partner network — also expanded with names like Fospha, Measured and INCRMNTAL, according to Unite.AI — shows this is intended as a full-fledged ad platform, not a simple test.
Early results — reported by OpenAI itself
OpenAI says an early analysis of global campaigns optimized for deeper-funnel outcomes showed that 52.7 percent of eligible one-day view-through conversions occurred within one hour of the matched ad impression. The company also says it is already seeing conversions after ad impressions even without qualifying clicks.
Among partner-reported results: according to DV Rockerbox, WeightWatchers saw a 15.3 percent lower attributed CPA than a blended paid-search benchmark, and according to Triple Whale, 93 percent of Portland Leather's visitors from ChatGPT Ads were new. All of these figures come from OpenAI and its partners and are not independently verified.
Safety and privacy — promises so far
OpenAI says the safeguards are designed to prevent ads from appearing in emotionally vulnerable, sensitive or otherwise inappropriate conversations, and that measurement partners will not get access to private user conversations. BleepingComputer also notes that the company says the ads will not affect ChatGPT's responses.
These points are company assurances. How they will be enforced — and what DoubleVerify and IAS will actually evaluate in the pilots — has not yet been described, and there is currently no independent review of the claims.
Open questions
- The advertisers: the list of the initial test group has not been disclosed.
- Pricing: no price or bidding model is known.
- Paying users: it is unclear whether subscribers will also see ads, or whether the format will remain free-tier based (noted by AI Weekly).
- Timeline: the exact launch date in October has not been specified.
- International rollout: no plans announced beyond the US test.
- Verification: all coverage traces back to OpenAI's announcement and BleepingComputer's report — no independent confirmation exists yet.