Confused when ChatGPT displays ads

OpenAI last week announced it would include ads in the free and Go plans, not showing up in more expensive plans like Pro, Plus, Business and Enterprise. Ads are scheduled to appear at the end of the conversation, based on the topic discussed. Users have some control such as turning off ads, seeing why a particular ad is being shown, turning off personalization to disable targeting.

OpenAI affirms that it does not show ads to user accounts under 18 years old and does not sell data to advertisers. The company affirms that ChatGPT still maintains “answer independence”, meaning ads will not affect the answers the chatbot provides.

However, the move still makes many people worried. One of the key questions is whether OpenAI is ready to follow through on its commitments as advertising becomes central to its business.

The decision is not surprising

15 years ago, social networking platforms struggled to find ways to profit from their huge user base.

The breakthrough came thanks to personalized advertising, in which ads are targeted to what users search for and pay attention to. This model became the main source of revenue for Google and Facebook, reshaping their services to optimize interactions with users.

Large-scale AI is extremely expensive today. Training and operating advanced models requires massive data centers, specialized semiconductor chips, and endless engineering resources. Despite rapid user growth, many AI companies are still incurring losses, with OpenAI alone expected to cost $115 billion in the next five years.

Only a few businesses can afford such costs. Most AI vendors desperately need scalable revenue models. Targeted advertising is the most obvious answer, as it is still the most stable way to profit from a large number of users.

OpenAI’s promises in the past

OpenAI’s commitment to maintaining answer independence and protecting privacy may reassure many users, but they are still based on vague statements that are open to different interpretations.

The company states it “doesn’t show ads near sensitive content like health, mental health, and politics,” but doesn’t say what’s considered “sensitive,” a clear definition of “health,” or who decides the limits of this content.

Much of the conversation with AI in the real world is outside such very narrow domains. OpenAI has not provided details on what ad content will or will not be displayed.

“Without restrictions on ad content, it’s easy to imagine a user asking ‘how to unwind after a stressful day’ and seeing an ad for an alcohol supplier. A question about ‘weekend entertainment ideas’ could lead to an ad about gambling,” author Raffaele Ciriello wrote on The Conversation.

These are all related to factors that are harmful to health and society. When juxtaposed with personalized guidance at the moment of decision-making, such ads can have a silent but powerful impact on user behavior, even when they do not address specific health issues.

Early social networks made similar promises about user protection. However, history shows that economic pressure will put pressure on the limits set by companies, causing them to chase profits and leaving users exposed to many threats.

Many advertising ideas have had a negative impact on social benefits. In 2018, the Cambridge Analytica scandal showed that personal data collected for advertising is at risk of being misused for political influence. Meta knows that its platforms can cause serious damage, including devastating mental health, for young people, but still refuses to make changes that would hinder advertising revenue.

The risk is greater with chatbots

Chatbots are more than just social media message boards. Many people use them in very personal ways to get advice and psychological support. Discussions with chatbots are often private and non-judgmental, filled with information that users would never make public.

This belief has an impact far beyond social media interactions. Users seek support and make important decisions when consulting chatbots. Even when separate from the chatbot’s responses, ads will still appear in private conversations. Messages about products, lifestyle choices, economics and politics juxtaposed with personalized guidance will have an impact far beyond ads while surfing social networks or websites.

OpenAI considers ChatGPT a “super assistant” for everything from economics to health, making the line between objective advice and persuading users in a certain direction increasingly blurred.

The core issue is conflict of interests. Advertising models often promote maximizing user interaction, but the content that maintains the most attention is often fake or untrue news, plays on psychology or is harmful to health.

“This is the reason that self-imposed barriers by online platforms have repeatedly failed,” Ciriello said.

Meanwhile, at the World Economic Forum (WEF 2026) taking place in Davos on January 19-23, Hagemann Snabe, Chairman of Siemens, also mentioned this issue when discussing artificial intelligence law. “The most important law to ensure AI adheres to human values ​​is to ban advertising-based business,” he said. “Advertising AI models will push companies to optimize interactions, leading to negative consequences for mental health – the problem already seen with social networks, will be much worse.”