OpenAI Sparks an Ads War, Stirs Privacy Concerns
The ChatGPT maker is testing integrated advertising
Ads are coming to AI, or more precisely, to ChatGPT. That’s what Anthropic, the company behind rival chatbot Claude, not-so-subtly proclaimed Sunday.
In a slate of Super Bowl commercials, Anthropic poked fun at its competitor’s foray into advertising. One, featuring a young man seeking advice on achieving a pull-up, was promptly directed to purchase “StepBoost Max” insoles by a bystander with a chatbot-like voice.
The campaign encapsulates what some fear could happen when exchanges with AI are commercialized: that the experience will be reduced to a game of dollars and cents, at the expense of utility.
Until recently, ad integration wasn’t in the cards for ChatGPT maker OpenAI. Chief executive Sam Altman had long railed against the move, once referring to it as a “last resort.” Altman has since changed his tune.
Ads started appearing Monday in the free and low-cost tiers of ChatGPT across the U.S.
The world’s leading chatbot draws more than 800 million weekly active users, according to OpenAI. Most don’t pay a penny for the service—but the company does. Exact financials are unclear, but the most recent estimates suggest OpenAI spends more than $700,000 per day servicing billions of ChatGPT queries.
As OpenAI races toward a fourth-quarter initial public offering, the pressure is on to prove commercial viability. The company burned through roughly $9 billion in 2025, and costs are mounting. Advertising revenue may help offset sky-high infrastructure and computing outlays, but it may also present new challenges.
Safety and transparency
OpenAI was born in 2015 as a nonprofit research lab, committed, in company parlance, to “ensur[ing] that artificial general intelligence benefits all of humanity.”
The lab has since undergone some restructuring. It now maintains a nonprofit arm and a for-profit subsidiary.
As the company has leaned into commercial endeavors, it has courted government and enterprise clients, who pay handsomely for its services. But OpenAI still isn’t profitable and doesn’t expect to be until the end of the decade, unless something substantive changes.
Startups with burn rates like OpenAI’s can’t rely exclusively on subscriptions. Like its social media and search-engine forebearers, it will need to diversify its revenue streams.
It was less a question of whether OpenAI would bring ads into the mix and more a matter of when, for S. Shyam Sundar, co-director of the Media Effects Research Lab at Pennsylvania State University.
“I’m not surprised at all—in fact, I’m surprised it’s taken this long,” Sundar said.
The business model across the tech industry has long been ad-driven. Platforms such as Google and YouTube don’t require subscriptions to use their services; instead, they rely on sponsored content to generate revenue. Users supply their data and attention in exchange for access to the web.
But ads delivered through conversational AI will be different, Sundar said. Most people, when they use a traditional search engine, engage in what academics call “banner blindness,” the tendency to ignore pop-up content on a website’s periphery. In an AI-mediated chat, users won’t be able to filter out unwelcome commercial material; it will simply be part of their dialogue with the bot, according to Sundar.
The more challenging it is to avoid ads, the more persuasive they can become, he said. User autonomy is paramount if OpenAI and its advertisers want to engender trust and establish brand loyalty, he added.
In a study he conducted last year, Sundar and his co-author found that providing users the ability to opt out of ads—by prompting the bot not to display them—established greater trust, as did the ability to customize their ad experience. The study is currently under peer review.
“It’s entirely okay to be able to, sort of, take the stigma out of advertising,” Sundar said. “But you have to democratize it in some way whereby the end users have a say in the nature of the advertisement.”
A lack of autonomy might pose risks for those who have turned to AI as a confidante, Sundar said. People are increasingly sharing secrets, along with sensitive medical and financial information, with chatbots. OpenAI says such information won’t be sold to advertisers and that ads won’t influence ChatGPT responses.
The company has also pledged to keep ads separate from organic answers and to label them clearly. Users need to be able to trust that the bot is supplying useful, objective information, OpenAI wrote in a recent blog post.
But simply outlining these commitments isn’t enough, a former OpenAI researcher recently wrote in a New York Times opinion piece. The company should be subject to independent oversight to ensure it upholds its stated aims, she wrote.
Internet safety experts broadly agree.
There are several examples of mission-driven social platforms that, after going public, abandoned the values that once anchored their businesses, said Miranda Bogen, who directs the AI Governance Lab at the Center for Democracy & Technology, a nonprofit.
When it switched to an ads-based business model in the early 2000s, Meta—then called Facebook—received intense consumer pushback. Users alleged the company’s central priority had shifted from providing connection to monetizing attention.
Bogen said there’s no reason to believe this time, with AI, will be different.
“While things are being positioned as simply a sort of pilot at the moment, I think there are so many examples of exactly the path that this will go down,” she said.
Early days
One key hire may have suggested ads were the path forward long before OpenAI announced the strategy.
The company’s decision to install Fidji Simo as head of applications made things clear for Brian Wieser, CEO and principal at Madison and Wall, a consulting firm. Simo is known for her experience monetizing ad platforms at Meta and at Instacart, where she previously served as chief executive.
“You often have to give more consideration to what companies do rather than what they say,” Wieser said.
It was also foreseeable that even a well-capitalized startup such as OpenAI would need to add more revenue channels in order to scale, he added.
OpenAI’s biggest competitor, Anthropic, has taken a different tack. The lab has emphasized B2B relationships and has vowed not to feature ads in its products, including its flagship chatbot, Claude.
That appears to be paying off for Anthropic, which is pulling ahead of OpenAI in the sprint to profitability, a winning streak buoyed by the rapid uptake of its coding software among enterprise clients.
Unlike Anthropic, OpenAI has positioned ad placement as a helpful perk for its users. The company provided a use case for personalized ads in a recent press release. Someone attending a potluck asks ChatGPT for help finding recipes. Then, the bot connects them with vendors from which to purchase ingredients.
But in more serious cases, when people come to ChatGPT looking for guidance and support, shopping recommendations may not be appropriate. A chatbot could easily misinterpret a prompt, serving unwanted and unhelpful ads, said César Zamudio, an associate professor of marketing at Virginia Commonwealth University.
Several of OpenAI’s competitors, including Perplexity and Microsoft Copilot, have tested sponsored content. But Google’s Gemini, a tool more analogous to ChatGPT, remains ad-free, raising the risk that OpenAI’s user base may jump ship, Zamudio said.
There are still plenty of unanswered questions as ads enter conversational AI, according to Jennifer King, a research fellow at the Stanford Institute for Human-Centered Artificial Intelligence. Those outside the company have limited visibility into how OpenAI—or any other lab—stores the data it collects and, more importantly, how that data is deployed, King said.
If nobody’s minding the switches, the most intimate details shared with chatbots may become fodder for advertisers, undermining the lab’s mission and scaring off users, according to King.
“People may want to know that they can ask questions and have conversations in general that end up not being commercialized,” King said. “So with all these things, how [OpenAI does] it really matters.”