ChatGPT sells out
OpenAI tests ads in ChatGPT at premium rates, while Anthropic markets Claude as ad-free, igniting debate over AI trust and sponsored answers

photo by Mike Kemp / In Pictures via Getty Images
A version of this article originally appeared in Quartz’s AI & Tech newsletter. Sign up here to get the latest AI & tech news, analysis and insights straight to your inbox.
ChatGPT finally joined much of the rest of the internet and started running ads. On Feb. 9, OpenAI began testing sponsored placements at the bottom of answers for free-tier users.
OpenAI is asking some brands to commit at least $200,000 and charging $60 per thousand ad views, a rate that puts it in the same bracket as premium TV and streaming. A typical banner ad on the open web costs $2 to $10. OpenAI is pricing itself alongside the Super Bowl, not the sidebar.
Companies are biting: Williams-Sonoma, Target $TGT, Adobe $ADBE, and Audible have all reportedly signed on.
But this isn't search advertising with a chatbot skin. It's something fundamentally different, and the distinction matters.
In traditional search, users type a query and receive a list of options. The ad sits beside or above organic results, and the user understands they are comparison shopping. The power dynamic is clear. In a chatbot conversation, the dynamic shifts.
The user is asking for help, often about something personal, sometimes something vulnerable. The AI responds in natural language, as a single authoritative voice rather than a buffet of links.
When a sponsored placement appears at the bottom of that answer, it occupies a different psychological space than a Google $GOOGL Shopping ad ever did. It's the difference between a billboard on the highway and a suggestion from your friend.
Anthropic, the company behind Claude, has leaned hard into this distinction. Its four Super Bowl ads depicted absurd scenarios where a chatbot steered users from genuine questions toward sponsored products.
A guy asking how to talk to his mom gets pointed to a dating site. Someone seeking fitness advice is sold height-boosting insoles. The ads were funny.
They were also, according to OpenAI CEO Sam Altman, "clearly dishonest." OpenAI's policy says sponsored content won't appear near sensitive topics like health or mental health and won't influence the AI's answers.
But Anthropic is tapping into a real anxiety about a technology people are still figuring out how to trust, and it appears to be resonating. In the days after the Super Bowl, Claude's app climbed from No. 41 to No. 7 on the US App Store, with downloads jumping 32%. The "no ads" pitch, it turns out, is itself a potent form of marketing.
The real tension isn't whether ads will literally hijack AI responses today. It's whether the incentive structure that advertising creates will subtly reshape what these products become over time. Anthropic's published position is that even well-behaved ads introduce pressure to optimize for engagement, and that engagement metrics don't necessarily align with being genuinely useful.
That's a familiar worry from social media's evolution, now arriving in a new context where the conversational format makes the stakes feel more intimate.
Follow the money, find the future
Not everyone in the AI industry agrees on which way to jump.
Perplexity, the AI search startup that was among the first to introduce ads into its product, has started pulling back. Executives recently said at a media roundtable that ads risk making users "start doubting everything," and the company is pivoting toward enterprise sales and subscriptions instead.
Google, meanwhile, is testing ads in its AI Mode search results and has signaled that Gemini will eventually carry advertising too. Microsoft $MSFT Copilot has had sponsored content baked in from the beginning. Grok, the chatbot built into Elon Musk's X $TWTR, is expected to follow the same ad-supported playbook.
It’s not surprising that other companies are considering ads. Training and running these models costs billions. OpenAI spent an estimated $5 billion more than it earned last year. Venture capital has kept the industry afloat so far, but investors want returns, and free tiers don't generate them. Ads are the most proven monetization engine the internet has ever produced. It would be stranger if OpenAI didn't try them.
There may be a third path for brands, at least. Airbnb $ABNB CEO Brian Chesky recently told investors that traffic arriving from AI chatbots converts at higher rates than traffic from Google. He didn't share specific numbers, but called chatbots “really good top-of-funnel discoveries.”
If that holds, companies may not need to buy ads inside the conversation at all. They just need the AI to recommend them. That's a promising signal for advertisers still weighing whether ChatGPT's $60 price tag is worth it. But it does nothing to solve OpenAI's problem.
Running these models is extraordinarily expensive, and someone has to pay for the free tier. If not advertisers, then who?