OpenAI starts laying foundations for ChatGPT ads in EU

Advertisers can flag or exclude specific actions — a purchase, say, or a sign-up — from tracking. Moreover, the update also adds a country field to the data the pixel collects, suggesting OpenAI is building it with jurisdictional data handling in mind. 

None of that is required for advertising in the U.S., where privacy law operates on an opt-out basis — users can request that tracking stops but businesses don’t need permission before it starts. In Europe, it works the other way: explicit consent is required upfront before a tracking pixel can fire and users must be able to withdraw it at any time. 

“OpenAI’s move into advertising looks like it’s being built with European regulation in mind, and it has to be, with governments increasingly focused on the next phase of the platforms’ societal impact as we enter the AI era,” said Enthropy Consulting’s founder, Alex Tait. 

That scrutiny is sharpened by a public that is more attuned than ever to the machinery built around monetising their attention.

“A consent signal has to travel cleanly through every partner in the advertiser’s tech stack —  and that transmission problem is where ad tech has struggled for years,” said Sourcepoint COO Brian Kane. “Each handoff creates a potential compliance risk, one regulators and privacy groups will be watching closely.”

Signs that a pixel was on the way first emerged last month. It reflects a broader strategic choice born out of necessity. Third-party cookies — the small files that advertisers have long used to track what users do after clicking an ad — have been blocked by Safari and Firefox for years, making it harder for advertisers to measure whether their campaigns are actually working.

Google and Meta have spent considerable effort building server-to-server alternatives in response, where data passes directly between systems rather than through a browser. OpenAI simp;ly skipped that chapter, building that way from the start. 

That said, the conversion pixel is still maturing. Early advertisers in ChatGPT’s ads pilot have no pixel at all. They were manually tracking how much traffic their ads drove using rough methods and spreadsheets. The pixel is still being rolled out, and in its current form only measures the last action a user takes before converting such as clicking on an ad. More sophisticated measurement, like crediting ads that a user saw but didn’t click or adjusting how far back in time a conversion can be attributed — is planned but without a confirmed timeline. 

Until then, the pixel is deployed more like a managed service than a self-serve product. OpenAI builds it for advertisers based on what they want to measure and assigns someone to help with implementation. That will need to change if the ads business is to scale. 

“Now that OpenAI has got the conversion pixel, they’re going to start getting those implemented,” said Jai Amin, chief solutions officer, media activation at Jellyfish. “I’d imagine the narrative around this just being brand budgets is going to change once they start to see the data and conversion sales information coming in.”

As things stand, OpenAI’s ChatGPT ad pilot is extending beyond the U.S. and will soon include Canada, Australia and New Zealand in the coming weeks. Digiday also reported last week that the company is looking for execs for its ads team in London and Tokyo.

OpenAI did not respond to Digiday’s request for comment.