OpenAI pulls ChatGPT's ad-like app promos after backlash

You finish asking ChatGPT to debug a gnarly BitLocker issue, and right underneath the answer, a friendly prompt suggests you book a Peloton workout.

That awkward reality started showing up on screens over the weekend, sparking a massive backlash from users who felt OpenAI had finally slipped ads into the chatbot. Now, the company is hastily pulling the plug.

The Peloton pitch nobody asked for

OpenAI Chief Research Officer Mark Chen took to X to confirm the feature has been turned off. He admitted that anything feeling like an ad requires extreme care, and conceded that the team simply fell short this time around.

Screenshots flooded social media showing integrated partner cards for brands like Target and Peloton sitting beneath completely unrelated conversations, such as discussions about xAI. What made folks especially furious was that paying Plus and Pro subscribers were seeing them too.

Behind the scenes, OpenAI seemed confused about its own messaging. Data engineer Daniel McAuley insisted the prompts were not technically ads because there was no financial component, though he admitted the utter lack of relevance made for a confusing experience.

Meanwhile, ChatGPT head Nick Turley pushed back entirely, claiming there are no live ad tests and that viral screenshots were either fake or not ads. He stopped short of clarifying which specific images were supposedly fabricated.

Billion-dollar burn versus user trust

It is no secret why OpenAI is itching to find new revenue streams.

While annualized revenue touched $12 billion this summer, projections show the firm could burn $115 billion through 2029 on its quest toward superintelligence. The Financial Times recently reported that only about 5% of ChatGPT's 800 million active users actually pay a subscription.

Both Sam Altman and Nick Turley have previously noted they are not fundamentally opposed to ads, so long as they feel tasteful like Instagram. But with Anthropic and Google breathing down its neck, OpenAI reportedly declared an internal code red last week, putting shopping and ad experiments on the backburner to focus strictly on model quality.

Why it matters

We think this misstep shows just how tricky monetization is going to be for conversational AI.

In traditional search, slapping sponsored links at the top works because users expect an index of the web. But in a one-on-one conversational interface, having your assistant suddenly pivot from cybersecurity to grocery runs feels invasive and breaks the illusion of unbiased advice.

Pushing these promos onto paying Pro subscribers was an especially tone-deaf unforced error. Telling users it is not an ad because no money changed hands is pure tech pedantry that completely misses how people experience software.

Google is already testing ads inside its Search AI Mode, meaning commercialization in AI is inevitable. OpenAI was smart to shut this experiment down quickly, but finding a way to bankroll its massive compute bills without alienating its base remains an unsolved puzzle.