Copy article

OpenAI's Ads U-Turn Fiasco

ended 08. December 2025

OpenAI has disabled ChatGPT's app suggestion feature following complaints from paying subscribers who reported seeing promotional messages for companies including Peloton and Target during conversations.

The company's chief research officer Mark Chen acknowledged on 6 December that OpenAI "fell short" with the promotional messages and confirmed the feature has been turned off whilst the company improves precision. Chen stated OpenAI is "looking at better controls so you can dial this down or off if you don't find it helpful."

Multiple ChatGPT Plus and Pro subscribers posted screenshots on social media showing app suggestions appearing during unrelated conversations. One user reported seeing Target shopping recommendations whilst asking about Windows BitLocker. Another reported Peloton recommendations appearing during a conversation about Elon Musk.

OpenAI executives maintained there are no advertisements or advertising tests currently live in ChatGPT. The company stated it was testing ways to display apps built on the ChatGPT app platform announced in October, with "no financial component" to the suggestions.

ChatGPT head Nick Turley posted on 6 December: "There are no live tests for ads – any screenshots you've seen are either not real or not ads. If we do pursue ads, we'll take a thoughtful approach."

The Wall Street Journal reported this week that OpenAI CEO Sam Altman issued a "code red" memo prioritising improvements to ChatGPT user experience and output quality and pushing back other products including advertising.

We want your views:

  • If computing costs ($5 billion) exceed total revenue ($3.7 billion) before any other expenses, how can AI companies sustainably serve paying customers without advertising?
  • Does the phrase "no financial component" accurately describe situations where companies test commercial partnerships that could become paid placements?
  • What's the ethical boundary between "helpful recommendations" and revenue testing when the company's stated goal is reaching profitability?

4 responses from the Newspage community

Copy all

Star Quote
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
OpenAI spent $5bn running ChatGPT whilst making $3.7bn in revenue. That's not a rounding error, that's a business model built on venture capital fumes. So when £160/month Pro subscribers started seeing ads during 'technical troubleshooting', the maths suddenly made sense.

The company insists these weren't ads because 'no financial component' existed. That's nonsense. If you're testing which commercial partnerships users tolerate before monetising them, you're conducting advertising research on paying customers.

Here's the sustainable path: charge what the service actually costs to run. Get more power users on the highest tier. Get more freebie seekers paying. Make it easier to connect to the API version. What doesn't work is turning premium AI assistants into glorified shopping channels whilst insisting you're just 'being helpful'. OpenAI learned the hard way that trust breaks faster than code tests.

Altman is right on the code red. Fix the chronic black box trust issues first.
Copy

Editor
Ranald Mitchell
Director at bad credit mortgage specialists Charwin Mortgages
OpenAI’s pause on app suggestions, after users saw prompts for brands like Target and Peloton, highlights a bigger truth: running frontier AI costs far more than subscriptions can cover. Long-term sustainability will rely on enterprise revenue, efficiency gains and clearer pricing, not creeping commercialisation. OpenAI may be correct that tests had “no financial component,” but users want honesty about whether such features could become paid placements. When people pay to avoid ads, trust is everything. Helpful suggestions are fine; opaque monetisation experiments are not. Transparency must come first.
Copy

Rohit Parmar-Mistry
Founder at Pattrn Data
This reeks of the same arrogance we saw when Meta "tested" ads inside paid Oculus games like Blaston in 2021. They claimed it was a test; users rightly saw it as a betrayal. The backlash was swift, and it should be the same for OpenAI.

There is a fundamental social contract in tech: if the product is free, I am the product. If I pay, I expect to be the customer, not the merchandise. OpenAI calling these "helpful suggestions" with "no financial component" is corporate theatre. It’s a distinction without a difference. Whether you label it a "suggestion" or an "ad," you are interrupting my workflow to sell me something I didn't ask for.

The economics of AI are clearly terrifying if $5bn in compute costs can’t be covered by subscriptions alone. But breaking user trust to plug that gap is a short-term fix. If OpenAI wants to be taken seriously as enterprise infrastructure, they must stop treating professional users like a monetisation experiment.
Copy

Patricia McGirr
Founder at Repossession Rescue Network
The moment an AI slips a brand into your chat uninvited, every claim about user trust starts to look shaky. OpenAI can apologise for glitches, but the numbers are the real story. When compute costs tower over revenue, the pressure to commercialise the conversation is built in. “No financial component” doesn’t reassure anyone when customers are already seeing Peloton in a thread about BitLocker. This is the bind every AI firm now faces. You cannot keep a platform clean and lean when the business model is anything but.