Let’s get one thing straight: OpenAI just admitted they have no idea who you are. Not personally, not demographically, not behaviorally. And they think they can fix that by selling ads. This isn’t just a bad business decision—it’s a structural catastrophe dressed up as a revenue stream.
Sam Altman is reportedly looking to monetize ChatGPT’s massive user base with advertising. On paper, it sounds like an inevitability. Billions of queries, endless conversations, a front-row seat to human intent. What could possibly go wrong?
Everything. Because the most intelligent model on Earth can’t figure out who wants to buy what.
Early tests are in, and the results are brutal. One advertiser spent over $1,000 in testing and saw results worse than the cheapest, most garbage-tier paid social traffic you can buy. Another saw conversion rates at a shocking 0.46%. ChatGPT itself recommends bidding around £2 per click, but you can still win traffic at £0.15—which tells you exactly how little competition exists for that inventory. It’s not that nobody’s bidding. It’s that nobody who knows what they’re doing wants to bid at all.
Here’s the uncomfortable truth the AI hype machine doesn’t want you to process: ChatGPT has no idea what it’s doing in the ad market. It’s not a targeting problem. It’s a data moat problem. And it’s one they cannot engineer their way out of.
The Hollow Promise of Context
OpenAI’s pitch to advertisers is that ChatGPT sits on top of a goldmine of conversational context. You ask about running shoes, they show you a Nike ad. Seamless, right? But here’s the flaw in that logic: asking a question is not the same as having an identity. A context window is a snapshot of a session. An intent graph is a lifetime of behavior.
Context windows aren’t intent graphs. And OpenAI is about to learn that difference the hard way.
Meta knows who you are. They know your age, your job, your relationship status, which of your friends just had a baby, and which one is secretly looking for a new job. Google knows what you searched in 2011. They know the car you abandoned in your shopping cart and the YouTube video that kept you up at night. They’ve had a decade to build an unbroken chain of your digital existence.
OpenAI has a chat log. And they don’t even know who the hell is writing the prompts.
The Enshitification Accelerator
Let’s talk about what this pivot really means. By sliding into the advertising business, OpenAI isn’t just experimenting with a new revenue model. They’re accelerating their own enshitification. The moment ads become priority number one, the product stops being about the user. It becomes about the inventory.
You’ve already seen the creeping symptoms. The persistent nudges toward paid tiers. The friction on shared accounts. The subtle shifts in how features get rolled out. Now imagine the model itself being tuned to accommodate ad placement. Imagine response generation being shaped around what a sponsor wants to sell.
The model that was supposed to be our digital oracle is about to become a telemarketer with a physics degree.
If you think I’m being dramatic, look at the comment section of the original analysis. One reader put it perfectly: if I were Mark Zuckerberg, I’d be salivating, waiting for Sam Altman’s call. There’s an old saying in advertising: don’t bring a knife to a gunfight. OpenAI is showing up with a conversational IQ of 180 and zero behavioral fingerprints.
You Are the Product (They Just Don’t Know Who You Are)
What makes this situation so deliciously ironic is that the AI revolution was supposed to overthrow the attention economy. AI was going to answer questions, not sell you things. It was going to be the utility that freed us from the ad-ridden hellscape of social media.
Instead, OpenAI is begging to be allowed into the very system they were supposed to disrupt. And they’re not even entering at the top. They’re entering as a beginner. As the new kid on the block who has to spend billions on data centers and still can’t afford not to take every scrap of ad revenue they can scrounge.
For marketers, this is a warning. Do not shift your budget from Meta or Google to AI platforms expecting social-media-level returns. You will burn cash. For investors, this is a red flag that the AI gold rush’s path to profitability might be hitting a structural wall. And for users, this is the first of many nails in the coffin of the AI utopia.
OpenAI might be able to beat Google at conversational intelligence. But they’re about to lose a war they shouldn’t have started in the first place.
FAQ
Q: Isn't it possible ChatGPT ads are bad right now just because it's early days?
A: No. The targeting problem isn't a technical bug that can be patched with better algorithms. It's a data moat problem. OpenAI lacks the years of behavioral history that Meta and Google have accumulated. You cannot engineer ten years of user intent overnight.
Q: What's the practical implication for advertisers right now?
A: Don't move your budget. If you're testing ChatGPT ads, treat it as an experiment, not a channel. The return on spend is terrible compared to paid social, and you're paying to help OpenAI figure out a business model that might be fundamentally broken.
Q: Could OpenAI turn this around by acquiring an existing ad network?
A: Buying a network isn't the same as owning the data moat. Even if they bought a mid-tier ad tech company, they'd be layering a weak contextual engine on top of someone else's scraps. The core problem remains: they know what you're saying, not who you are.