OpenAI has had ads in ChatGPT since February. It did not announce them. Now it has a partner that wants to build ads that talk back.
Smartly, a 13-year-old adtech company led by industry veteran Laura Desmond, signed a deal with OpenAI on Tuesday to become its first creative adtech partner. The immediate work is narrow: Smartly will help a select group of entertainment, retail, and sports clients tweak their ChatGPT ad placements in real time, building on the basic contextual ads OpenAI began running in February. The vision is larger. Smartly wants to build interactive, conversational ad formats inside ChatGPT — ads that use the same interface as the model itself, asking follow-up questions and responding to user input the way a chatbot does.
The Boots example Smartly cites is instructive. The UK retailer ran a Meta chatbot ad that opened a conversation window, asked about gift preferences, and served recommendations in response. Smartly said the format drove nearly five times as many sales as Meta's standard ad units. That is the template: bring the frictionless dialogue mechanic into a paid context, and measure the difference.
OpenAI's current ad rollout is basic and contextual. A user comparing smartphones might see a Best Buy ad. A traveler researching destinations might be shown an Expedia placement. Sensor Tower found more than 100 brands had advertised on ChatGPT within the first few weeks, with retail accounting for 44 percent of them. That is the baseline Smartly is being asked to improve.
The tension Smartly acknowledges but does not resolve is the value exchange question. Laura Desmond's framing — that people want to be known, that relevant ads are a service rather than an intrusion — is the industry's standard argument for behavioral targeting. It is also the argument that produced the cookie ecosystem, which regulators in Europe and California spent a decade dismantling. ChatGPT ads have not faced that scrutiny yet, partly because they do not exist at scale, and partly because the conversational format Smartly is proposing does not yet exist.
The timing matters. OpenAI is running at roughly $2 billion in monthly revenue, burning cash at a pace that makes every new monetization lever worth pulling. Ads in a chatbot are not novel — Google, Meta, and Bing all monetize search-adjacent conversations. But ads in a chatbot where the model generates the response stream in real time introduce a structural question that has not been fully answered: who is actually responsible for what the ad says, and who is liable when a conversational ad recommends something harmful or inaccurate?
The question extends beyond any single brand. OpenAI has a $2 billion monthly revenue run rate and is not yet profitable. Every monetization lever it pulls — subscriptions, API, ads — is a bet that the revenue growth justifies the infrastructure costs. Ads in a generative AI context are not a solved problem. They are an experiment that OpenAI is running in public, with users as the test group, because that is the only way to generate the data that will tell it whether the format works.
Smartly is not answering the hard questions. Its job is to make the ads work better, and to make the case that better is inevitable. Whether that case survives contact with how people actually respond to being sold to inside a chatbot is a separate question.
Sources: Business Insider | Business Insider - ChatGPT ads initial rollout | Business Insider - Sensor Tower data
