ChatGPT’s new image ads check your mood first

OpenAI will begin testing picture ads inside ChatGPT’s image generator later this month, starting in the U.S. with an initial group of advertisers, the company said Monday. The ads sit alongside the images people ask ChatGPT to make, and OpenAI says they’ll be clearly labeled and won’t influence ChatGPT’s answers, TechCrunch reported.

Until now, ChatGPT ads were a company’s name, logo and link in a “sponsored” section beneath the chat, The Verge notes. OpenAI says the pictures will help people imagine how a product “could fit into their lives,” according to Engadget. Paying for Plus, Pro or Enterprise turns the ads off entirely, and TechCrunch suggests that’s partly the point: the ads could push casual users to pay.

OpenAI says guardrails determine when a conversation is “appropriate” for ads, with the goal of keeping them out of “emotionally vulnerable, sensitive, or otherwise unsuitable contexts.” A chat about your personal health or mental wellbeing won’t get one, Engadget reports.

For each conversation, a system has to decide, before any ad appears, whether the person on the other end is in a state to be sold something. OpenAI presents that as protection, and for someone typing through a bad night it is. It’s also the signal an ad business would want most: a running read on which users are calm enough to pitch.

ChatGPT’s ads reached a pace of more than $1 billion a year in under 200 days, The Next Web reports, and OpenAI counts 1.2 billion weekly users. Alongside the image ads, OpenAI is adding measurement partners such as AppsFlyer and Adjust, tools advertisers use to tie a click or a sale back to the ad that caused it.

In Europe, ChatGPT ads have run in 31 countries since late August, for free users and those on the Go plan. On Aug. 31 the European Commission designated ChatGPT a very large online search engine under the Digital Services Act, which adds duties once a service passes 45 million monthly users in the EU. ChatGPT has 159.1 million. Starting in January, OpenAI has to keep a public archive holding a year of its ads, with the advertiser, how long each ran, who it was aimed at and how many people saw it.

That designation also lists risks to users’ mental well-being among the things OpenAI must assess. The DSA bans ads targeted using sensitive personal data, but The Next Web notes that rule is written for online platforms. The Commission classified ChatGPT as a search engine on purpose, to avoid handing OpenAI a legal shield over what ChatGPT itself writes.

OpenAI is also running brand-suitability pilots with DoubleVerify and Integral Ad Science, tests meant to show advertisers what their ads would appear next to. According to The Next Web, the two firms will evaluate how OpenAI applies its brand-safety standards in a controlled environment and without access to real conversations, so the outside check on where ChatGPT’s ads land happens with no real chat in view.

Sources

• TechCrunch: OpenAI launches visual ads that appear alongside image generation results
• The Verge: OpenAI is sticking more ads in ChatGPT
• Engadget: OpenAI is testing more ads in ChatGPT
• The Next Web: OpenAI is putting ads inside ChatGPT’s image generation