Ads arrive on ChatGPT Free and Go plans this month

On August 15, OpenAI emailed ChatGPT users to say ads may start appearing on the Free and Go plans later this month, along with a Privacy Policy update covering how ads work and how users can control them. At first, ads will be picked from what you’re discussing right now. Your history won’t be used. The full email was posted on Ycombinator’s Hacker News.
Ads in ChatGPT aren’t new. OpenAI started testing them on February 9 with logged-in adult users in the U.S. on the Free and Go tiers, and by August 11 they were live in the United Kingdom, Mexico, Brazil, Japan and South Korea. Plus, Pro, Enterprise, Business and Education plans have no ads, and OpenAI says “ad-free options remain available.”
For now, targeting is contextual. An ad is matched to the subject of the conversation in front of you, with no profile built up over time. OpenAI says it uses “the current topic of your ChatGPT conversation” plus limited contextual information, such as your general location and device type. Past chats and memories, the facts ChatGPT keeps about you between sessions, are not used to pick these ads.
Personalization comes later, and only with consent. Before it starts, OpenAI says it will ask users whether they want “more relevant and personalised ads.” Those who agree will get ads based on information that “stays only on ChatGPT,” and the email’s examples are ads you’ve interacted with and “context from your chats.” You’ll be able to manage ad personalization in settings. No start date for that phase has been given.
Measurement is a separate track. OpenAI uses data on which ads you view or interact with to measure and improve how well they work. Advertisers get only overall performance numbers, such as total views or clicks. The email says personal details and conversations are not shared with advertisers, who have no access to “your chats, chat history, memories, or personal details.”
OpenAI also promises that every ad will be clearly labeled and visually separated from ChatGPT’s responses. On the question users are most likely to ask, the email has one line: “Ads do not influence the answers ChatGPT gives you.”
Up next
WMTrace won’t ship an Anthropic watermark detector
“No guessed provider secrets”: WMTrace, an open-source watermark forensics tool, ships no Anthropic or OpenAI detector until an official verification API or public spec exists. Every key in it is a published research key.
Related stories
- Ads inside ChatGPT hit $1B annualized run rate
- ChatGPT ads cross into 31 European countries
- ChatGPT Business Premium seats go out at $100 a month
- US online businesses can publish their own agent in ChatGPT
- OpenAI’s ChatGPT ads hit a $1B run rate in under 200 days
- Microsoft shares slid as Claude got Sheets and Whiteboards
Comments
No comments yet. Be the first.
Join the conversation
Sign in to leave a comment. It appears after moderation.

