OpenAI ties ChatGPT ads to purchase tracking through LiveRamp, Tealium and Hightouch

Conversion data tells an advertiser whether a ChatGPT ad led to a purchase or another conversion ^[9]. Counting only the firms named in the report, ten outside companies now work on ChatGPT advertising: three for conversion data, five for attribution and two for brand suitability ^[14].

For a privacy team, the open question is how a chat session gets linked to a record in an advertiser's systems. Measuring conversions requires that link ^[9]. The report does not say how a conversion is matched to a ChatGPT user. It also does not say whether OpenAI's no-access promise covers the conversion and attribution partners, or how advertisers and their creatives are screened ^[5].

Ad placement already depends on what a conversation is about. OpenAI says its safeguards are designed to keep ads out of emotionally vulnerable, sensitive or otherwise unsuitable conversations ^[6]. BleepingComputer notes that suitability matters more in ChatGPT because ads appear around conversations, where elsewhere they would sit on a traditional webpage ^[13].

The record does not show a malicious-ad risk. Brand suitability keeps an advertiser's ad away from content it does not want to appear next to ^[4]. Screening out ads that could harm the person who clicks them is a separate control. OpenAI's own description of the format covers labeling. "Initially, we'll test this new ad format during image generation in ChatGPT. Ads will be clearly labeled, and remain separate from the image being created," OpenAI said ^[7]. The company also repeated that ads do not influence ChatGPT's answers ^[12].

BleepingComputer describes advertising as a way for OpenAI to earn money from users who may not pay for a subscription ^[10]. It also reports that OpenAI expects ads to become a much larger part of ChatGPT ^[11]. I'd put consumer ChatGPT on a privacy team's inventory of ad-supported services with conversion tracking now. I would not give it a malvertising rating until there is an incident or a published screening policy to judge.