OpenAI is launching a pilot program to introduce advertisements to its generative artificial intelligence service ChatGPT.
The company announced on its website May 9 (local time) that it would begin testing advertisements. The ad test targets select users in the United States who use ChatGPT's free version and the low-cost subscription model called the Go plan. Paid subscribers, including those on student plans and Pro models, are excluded from the test. OpenAI stated it plans to "expand the ad test to more users over the coming weeks depending on test results."
ChatGPT advertisements will be generated based on users' ChatGPT usage history. The AI analyzes the context of ChatGPT interactions and recommends products and services tailored to individual users. If users consent to ad personalization settings, past conversation data may be used to generate advertisements. Conversely, users can hide ads or provide feedback on ad displays.
Advertising messages are generated separately from responses to user queries. Ad messages are marked with a "Sponsored" label and visually distinguished from regular responses. Information about which users see specific ads is not shared with advertisers. Advertisers only receive aggregate data such as impression counts and page visits driven by ChatGPT ads.
OpenAI announced last month that it would introduce advertising services. Following this, Anthropic, which operates ChatGPT competitor Claude, recently drew attention by running a Super Bowl advertisement that took aim at OpenAI's ad service.