AI as a Sales Funnel: Can ChatGPT Stay Trustworthy When It Starts Selling?

On September 29, OpenAI announced that ChatGPT will begin earning a commission on sales made through its platform. Users can now ask for a product, receive recommendations, and buy directly within the chat. OpenAI takes a share of each purchase.

ChatGPT’s move into e-commerce looks small on the surface but marks a deeper shift. The platform was built as a knowledge tool, one designed to assist, not to sell. Linking conversation with transaction changes that relationship. The more OpenAI earns from purchases, the more its role begins to resemble a salesperson rather than a neutral adviser. The shift raises a question that will shape the future of AI commerce. Can an artificial intelligence stay credible when it profits from what it recommends?

Every digital platform has faced this problem once it turned commercial. Google’s early success came from its claim of objectivity, ranking results based on relevance. Then paid listings arrived. The logic was simple: advertisers fund the service, users keep it free. Over time, though, users learned that relevance had a price. The trust that built Google’s power also began to erode. Amazon followed the same path. The best deals are rarely the first results anymore; paid placements and private-label brands dominate the screen. These are convenient for sellers, but not always for buyers.

ChatGPT now sits at a similar crossroads. Its appeal lies in the illusion of impartial intelligence. You ask, it reasons, it answers. The introduction of commission-based sales threatens that foundation. If a chatbot suggests a pair of headphones and earns more from one brand than another, how can a user know the difference between advice and advertising? The problem is not that OpenAI wants to make money; it is how that motive changes the user experience.

There is a fine balance between convenience and manipulation. Imagine asking ChatGPT for a skincare routine and receiving a curated list of products, all available to buy in one click. The simplicity is seductive. But what if the brands appearing first are those that paid a higher commission? The user cannot see the business logic behind the answer. Unlike a web search, there is no “ad” label, no visible hierarchy of bids. The risk is that trust, once lost, is hard to rebuild.

Some will argue that this system could help small merchants. Integrated AI shopping could open visibility to niche sellers that lack the budget to compete on Google or Amazon. If ChatGPT can genuinely match users with the best products, regardless of brand size, it could rebalance online retail. That is a fair point, but it assumes OpenAI will design its ranking model to prioritise relevance over commission. History suggests otherwise. When profit enters recommendation, neutrality fades.

The deeper issue is one of information integrity. AI already carries bias from its training data. Adding financial incentives adds another layer. Once a platform starts earning per sale, the commercial bias becomes systemic. Even without direct interference, model tuning could start favouring products or sellers that bring higher revenue. Transparency becomes the only defence, yet OpenAI has not clarified how it will disclose paid results or ranking logic.

For marketers, this development changes the landscape again. Brands will soon need to optimise for AI responses rather than search rankings. The key question is what “optimisation” will mean. If it simply becomes another bidding system, marketing budgets will shift from Google Ads to ChatGPT placements. The medium changes, the model does not. But if OpenAI builds a fairer, insight-led discovery channel, it could set a new standard for value-based commerce. That would require a governance framework that prioritises accuracy, user benefit, and disclosure over short-term gain.

Every platform that reaches scale faces the same tension between trust and transaction. Subscription models promise independence but limit growth. Advertising and commission models drive revenue but distort incentives. OpenAI is now walking that line. To succeed, it will need to keep transparency at the centre of its design. Users must know when a suggestion is commercial and when it is not. Sellers must compete on quality rather than commission.

If OpenAI manages that balance, ChatGPT could evolve into a new kind of marketplace—one where convenience does not erase credibility. If it fails, it will follow the same path as the search engines and marketplaces before it: useful, profitable, but no longer trusted.

In the end, the question is not whether AI can sell. It already can. The question is whether it can sell without losing the reason people listen to it in the first place.