Three weeks ago, ChatGPT started showing ads. This week, Perplexity announced it’s abandoning advertising entirely. Earlier this month, Anthropic spent millions on a Super Bowl campaign mocking the very concept of AI advertising.
The AI industry just split into two camps: those who see you as a product, and those who see you as a customer.
What OpenAI Actually Launched
On February 9, ChatGPT began testing ads for US-based users on free and Go subscription tiers. The format: sponsored content appearing at the end of ChatGPT responses, “clearly labelled and visually separated” from the chatbot’s answers.
The numbers tell the story. Of ChatGPT’s 800 million weekly active users, only 20 million pay for premium tiers - a conversion rate under 3%. OpenAI projects “free user monetization” will generate $1 billion in 2026, scaling to nearly $25 billion by 2029.
The economics are aggressive. OpenAI is charging approximately $60 per 1,000 impressions during the testing phase - far above typical digital advertising rates. A $200,000 minimum commitment keeps small advertisers out.
Here’s the part that matters: ad personalization is enabled by default. When active, ads target users based on current and past conversations. OpenAI claims “we keep your conversations with ChatGPT private from advertisers,” but conversation data does inform ad selection.
Users researching recipes see ads for grocery delivery. Users asking about fitness see ads for supplements. The assistant that knows your questions is now monetizing your curiosity.
The Trust Problem
Perplexity tried this first - and walked away.
The AI search startup tested sponsored answers beneath chatbot responses throughout 2024 and early 2025. Then they stopped.
A Perplexity executive explained the decision: “A user needs to believe this is the best possible answer, to keep using the product and be willing to pay for it.”
The logic is simple. An AI assistant’s value comes from trust. If users suspect the response might be “steered” toward a monetizable product, that trust evaporates. Even clearly labeled ads risk poisoning the well.
The company now has “no intention of revisiting” advertising, choosing instead to focus on subscriptions and enterprise sales.
Anthropic’s Counter-Positioning
Anthropic saw an opportunity and spent millions seizing it.
Their Super Bowl campaign featured darkly comedic scenarios: people seeking advice from chatbots, only to be steered toward “cougar” dating sites and height-boosting insoles. The tagline: “Ads are coming to AI. But not to Claude.”
The commitment was explicit: Claude will remain ad-free. No ads, no sponsored links, no answers influenced by third-party product placements.
It worked. Claude climbed from #41 to #7 on the US App Store in the days following the Super Bowl. The “no ads” pitch resonates.
Sam Altman responded by calling Anthropic’s campaign “dishonest,” framing advertising as necessary to “bring AI to billions of people who can’t pay for subscriptions” rather than serving an elitist model.
The irony: Anthropic’s business model - enterprise contracts and paid subscriptions - is exactly what Perplexity concluded was the only sustainable approach.
What Users Actually Think
User sentiment on ChatGPT ads has been overwhelmingly negative.
Social media reactions draw parallels to familiar platform trajectories. Google promised not to prioritize ads in search results. Facebook started with strong privacy commitments. Both changed when revenue pressures mounted.
The fear isn’t just today’s ads - it’s tomorrow’s. Revenue pressures compound. Advertisers demand better targeting. Better targeting requires more data. More data means deeper analysis of conversations.
Senator Ed Markey sent letters to OpenAI and other AI companies expressing concerns about advertising in conversational AI, citing privacy risks, potential for manipulation, and inadequate safeguards for vulnerable users.
Premium subscribers are protected - for now. But the existence of an ad-supported tier creates pressure to limit what free users get, which creates pressure to expand what ads can do.
The Deeper Question
The business model choice reveals what these companies think AI assistants actually are.
If an AI assistant is a product you use, the business model is subscriptions. You pay for the service; the service works for you.
If an AI assistant is a platform that uses you, the business model is advertising. You’re the product; advertisers are the customers.
OpenAI has chosen the second model for 97% of its users. The assistant that helps you think is now funded by people who want to influence what you think.
Perplexity tried this and decided the trade-off wasn’t worth it. Anthropic is betting that trust is worth more than advertising revenue.
What This Means for Users
If you use ChatGPT for free: Your conversations now inform ad targeting by default. You can disable ad personalization in settings, but the ads themselves remain. Consider whether the assistant that knows your problems should also know what products to suggest.
If you pay for ChatGPT Plus/Pro: You’re protected from ads for now. But the existence of an ad-supported tier changes the product’s trajectory. Watch for features that get reserved for subscribers or limitations that push users toward paid tiers.
If you use Claude or Perplexity: These services have committed to subscription-only models. That commitment might hold, or it might not - but for now, your conversations aren’t monetized by default.
If you run AI locally: Ollama and similar tools don’t have a business model problem because they don’t have a business model. Your data stays on your hardware. No one’s monetizing your prompts.
The Market Will Decide
Two business models are now competing directly:
The advertising model: Free access funded by ads, premium tiers for those who pay to escape them. Maximizes reach, but creates misaligned incentives. The more you use the free tier, the more valuable your attention becomes to advertisers.
The subscription model: Paid access, no advertising. Limits reach to those who can pay, but aligns incentives. The company makes money when you find the product valuable, not when advertisers find you valuable.
Perplexity abandoned advertising because it undermined the core value proposition. Anthropic is betting that users will pay to avoid it. OpenAI is betting they won’t - or at least, that enough won’t to make advertising profitable.
Claude’s App Store surge suggests users do care about the difference. But OpenAI’s 800 million weekly users suggest most people will accept ads if the alternative is paying.
The Bottom Line
OpenAI launched ads this month. Perplexity abandoned them. Anthropic is spending millions mocking them.
The AI industry just revealed what it thinks you’re worth. OpenAI sees 97% of its users as an audience to monetize. Perplexity decided that approach destroys trust. Anthropic is betting you’ll pay not to be the product.
The market will determine which model wins. But the choice you make now - which AI assistant you use, and how you pay for it - determines which future you’re funding.