On 5 October 2026 OpenAI gave ChatGPT a visual ad format, a full third-party measurement stack and brand-safety controls, pitched as measurable, brand-safe AI advertising. The early proof comes from OpenAI's own launch advertisers and the partners it chose. Filip argues the durable play for Australian operators is to earn organic presence in AI answers now and verify every measurable claim against their own back-end data.
A number you cannot rebuild against your own sales data is a marketing claim, not a measurement.
The Take: OpenAI just turned the surface it sold as the end of advertising into an advertising platform, with a visual ad slot, a third-party measurement stack and brand-safety controls bolted on. The pitch to advertisers is measurable, brand-safe AI advertising. My read for any Australian operator reading this: the early proof comes from OpenAI's own launch advertisers and the partners it chose to grade the work, so the durable play is to earn organic presence in AI answers now and treat every "measurable" claim as unproven until it matches your own back-end sales data.
What actually changed on 5 October 2026?
On 5 October 2026 OpenAI gave ChatGPT a visual ad format, a full attribution and measurement stack and brand-suitability controls, aimed at a product that already reaches 1.2 billion weekly users.
The visual format launches in the United States first and rolls out later this month, with eventual reach across ChatGPT's 1.2 billion weekly users. It appears alongside AI-generated images, inside the answer itself, where the user came to get a job done and did not come to shop.
This is not a technology story. It is a measurement story. The engine people trusted because it carried no ads is now the engine that sells to them, and the firms being asked to validate that shift are the same firms that stand to earn fees from it.
Why are Australian operators the measurement test subjects?
Start with the size of the home crowd. 13.8 million Australians now use ChatGPT, nearly four in five of the country's AI users, which is not an audience any serious operator can wait out.
The behaviour underneath that figure is the real signal. 12% of Australians now use AI as their primary search tool, up from 5% a year earlier, a doubling in twelve months that tells you where the first click of a purchase decision is heading.
Australians now use ChatGPT, nearly four in five of the nation's AI users
We have measured the Australian market directionally, and the pattern is plain: the answer box is quietly becoming the place the buying decision gets anchored, and it is doing so while the ad load is still light. That window is the whole opportunity. Once the slot fills with paid results, organic presence inside the answer stops being free and turns into a bidding war.
Here is the trust gap that should worry any seller leaning on these ads. Only 35% of Australians trust technology companies to handle their data for AI, which means the audience is both enormous and sceptical, a combination that punishes clumsy advertising harder than a forgiving channel ever would.
Can you trust the launch numbers?
OpenAI did not just ship a format. It expanded measurement with data partners such as Hightouch, Tealium and LiveRamp alongside attribution partners including AppsFlyer, Triple Whale, Adjust, Branch and Kochava. It also began brand-suitability pilots with DoubleVerify and Integral Ad Science that run without access to private conversations. That is a serious stack. It is also a stack OpenAI assembled and OpenAI chose.
Look at the proof it led with, and look at who is vouching for it.
Every row is a confident number from a named brand. Not one of them has been through an independent audit, and each is measured against a benchmark the advertiser picked. A 15.3% improvement over "a paid-search benchmark" means little until you know which benchmark, on which products, over which weeks. A result graded by the party that sold you the space is a sales document wearing the costume of a measurement.
What I would do about it
Earn the organic answer first. Get your brand, products and proof into the material AI systems quote, so you show up inside the answer before you have to pay to sit beside it. Australian search behaviour is shifting fast enough that this is the cheap window and it will not stay open.
Refuse any result you cannot reproduce. When a platform or partner hands you an incrementality or lift figure, ask for the raw method and rerun it against your own back-end data. If the two numbers do not meet in the middle, trust yours and discount theirs.
Run a holdout, not a dashboard. If you do test the ads, keep a clean control region that sees none of them, so you are measuring genuine lift rather than watching the platform mark its own homework.
Start tiny and buy learning, not reach. Treat the first spend as the price of a private read on whether this channel works for an Australian buyer, with a hard cap and a fixed end date written down before a dollar moves.
Watch the ad load like a hawk. The value of organic presence drops the day the slot fills. Check monthly how much of the answer is paid, because that figure is your countdown clock.
My call
Here is how to read all of this. We score launches like this directionally, from public figures and our own read of the Australian market, with full awareness that launch-advertiser numbers are a tiny, self-selected sample and that the data is only days old. The methodology behind that read is written up here.
My opinion, and I am clear that it is an opinion: the firms rushing to buy ChatGPT ads on the strength of three case studies are volunteering to fund the seller's learning curve. The operators who win the next two years in Australia will spend that time earning the answer rather than renting it, and will hold every "measurable" claim to the only test that counts, which is whether it turns up in the bank.
If you want the deeper read on how AI answers are reshaping Australian demand, start here.
Frequently asked questions
Are ChatGPT ads available in Australia yet?
Not at launch. The visual format rolls out in the United States first from later in October 2026, so Australian operators have a short window to prepare before it reaches them.
Are the launch performance numbers trustworthy?
Treat them as directional at best. They come from a handful of OpenAI's own launch advertisers, are measured against benchmarks those advertisers chose and have not been through an independent audit.
What should an Australian business do right now?
Earn organic presence inside AI answers while the ad load is light, and build the habit of checking every platform-reported result against your own sales data before you believe it.
Will organic presence in AI answers stay free?
Unlikely. As the paid slot fills, the space your brand now occupies for nothing becomes something you bid for, which is exactly why the preparation matters before the format arrives.