Why this matters
If you've been treating ChatGPT Ads as a small-scale test, this is the moment to reconsider. OpenAI isn't just expanding into new regions, it's shipping the unglamorous plumbing (audience tools, conversion matching, reporting granularity, workflow automation) that usually signals a platform is getting ready to compete for real media budget. None of this is flashy on its own. Together, it's the infrastructure advertisers typically wait for before shifting spend off Meta or Google.
What's actually changing
Geographic rollout. ChatGPT Ads are expanding into select markets across Europe, India, and the Middle East and North Africa. That's a meaningful jump from a US-centric footprint, and it means agencies running international accounts can start testing the channel at a much larger scale than before.
Custom audiences got a lot more flexible. Advertisers can now update an audience by editing individual members instead of tearing it down and starting over, which saves real time for anyone managing audiences on a rolling basis. OpenAI is also letting advertisers combine multiple types of identifiers within one audience build, and audiences can now scale past 5 million members. On top of that, exclusion-audience rules have loosened, size estimates are more precise, and Google Advertising ID is now a supported identifier.
Measurement is catching up to expectations. OpenAI's Measurement Pixel can now match on more hashed customer data points, such as phone numbers, names, location, and zip or postal codes. The Conversions API picks up support for additional identifier types as well as Android's ad ID. Better matching means more conversions actually get attributed to the campaigns that drove them, which is the baseline requirement for any platform trying to be treated as a serious performance channel rather than a curiosity.
Reporting gets a finer lens for product ads. Brands running carousel-style product ads can now see performance broken out per product card, both in Ads Manager and via the Insights API. Worth noting: those individual card views are tracked separately and don't count as billable impressions.
What's coming next
OpenAI says it plans to roll out a new optimization model that factors in conversions from both direct clicks and post-view activity. Campaigns using that model will pay per impression, with OpenAI's system handling delivery to chase whatever conversion goal the advertiser sets. That's a shift toward the kind of automated, outcome-based bidding that already runs most of paid social and search.
The plugin worth watching
Separately, OpenAI has introduced a management plugin that lets advertisers build, monitor, and adjust ChatGPT ad campaigns directly inside ChatGPT or Codex using plain-language prompts. It can pull together ad creative from a site URL or a written brief, spin up variations, flag issues with running campaigns, and suggest changes, though any recommended update still needs a human to sign off before it goes live.
The bigger picture
None of these updates is a splashy new ad unit. They're the less visible groundwork, targeting, measurement, reporting, and workflow tooling, that platforms need in place before advertisers will commit serious budget. OpenAI appears to be building that foundation deliberately rather than chasing a single headline feature.
What marketers should do now
If you run paid media or advise clients who do, this is a good time to run a small ChatGPT Ads test, especially if you have exposure to the newly added regions. Set up the Conversions API properly from day one so you're not retrofitting attribution later, and keep an eye on the upcoming conversion-optimization model since it changes how campaigns get billed and optimized.