OpenAI’s mockup of its new visual ad format, with a carousel ad from an invented brand appearing during image generation in ChatGPT.
OpenAI is testing a visual ad format during image generation in ChatGPT, placed by guardrails that assess whether a conversation is emotionally vulnerable or sensitive before an ad appears. Ads already run in 31 European countries, and ChatGPT’s designation as a very large online search engine means it must publish a public ad repository by January.
OpenAI is putting advertisements inside ChatGPT’s image generation. The system decides where they go by judging whether a conversation is emotionally vulnerable. In Europe it will have to publish a record of all of it by January.
The visual format shows products in use or the experiences they allow, is clearly labelled and stays separate from the image being made, OpenAI said. Testing starts later this month in the United States with an initial group of advertisers. Advertising does not influence the answers ChatGPT gives, the company said.
ChatGPT reaches 1.2 billion people a week, by OpenAI’s own count, which it calls the largest AI-native consumer platform in the world. Advertising lets it serve a bigger market than subscriptions alone, the company said.
Europe already has the ads, and has had them for six weeks.
They arrived in 31 European countries in late August, for free and Go users only. Plus, Pro and Enterprise subscribers do not see them.
The regulator arrived in the same week.
The Commission designated ChatGPT a very large online search engine on 31 August, on 159.1 million monthly users in the EU against a threshold of 45 million. ChatGPT Ads passed $1B annualised in under 200 days.
The duties start in January, four months after the notification.
They include a public advertisement repository holding a year of ads, with the advertiser, the runtime, the targeting parameters and the reach. The Commission chose search engine over platform deliberately, to avoid handing OpenAI a liability shield over its own output.
That choice has a side effect on advertising that nobody appears to have intended.
The Digital Services Act bars ads targeted using sensitive personal data, and that rule is written for providers of online platforms. ChatGPT is a search engine.
OpenAI’s placement guardrails assess whether a conversation is emotionally vulnerable, sensitive or otherwise unsuitable before an ad appears. The Commission’s designation named risks to users’ mental well-being among the things OpenAI must now assess.
DoubleVerify and Integral Ad Science will evaluate how the brand safety standards are applied, in a controlled environment and without access to real conversations. OpenAI says the ads are clearly labelled and stay distinct from its answers.
Get the TNW newsletter
Get the most important tech news in your inbox each week.
