OpenAI has made a predictable yet controversial move: ChatGPT, the AI tool millions rely on for health, finance, and more, will start displaying advertisements. The company presents this as a measure to "democratize AI," but its financial forecasts reveal another goal: they expect to reach $1 billion in advertising revenue by 2026 and scale up to $25 billion by 2029. This is not a trial but a new growth strategy. OpenAI claims that ads will not influence responses, that privacy will be protected, and that everything will be transparent. However, experience with digital advertising shows that the gap between promises and reality is often significant. When a company plans to generate billions by selling ad space in a conversational AI, the question is not whether there will be a conflict of interest, but when it will become evident. Advertising in AI is not just a simple banner: it is more subtle, integrated, and persuasive. The so-called "commercial bias" describes how a model may prioritize sponsored responses. If you ask for the best video editing software, the AI might elaborate on the sponsored option and barely mention free alternatives. OpenAI asserts that this will not happen, but the visual integration already creates confusion: where does objective information end and commercial suggestion begin? The search for information transforms into agentic marketing. Ads in ChatGPT not only showcase products but also allow direct actions: "Do you want me to book this table for you?" Practical, yes, but it raises the question: does it recommend that restaurant for quality or for commercial agreements? User segmentation is clear: in the free version and ChatGPT Go, the user is the product; their data and patterns feed monetization. In Plus and Pro, users pay to avoid that indirect exploitation. It’s the old internet model applied to AI: if you don’t pay, you are the product. This raises a dilemma beyond the commercial. If access to unbiased information depends on the ability to pay, a new inequality is created: a premium AI for those who can afford it and a sponsored AI for the rest. Thus, a new battleground emerges: GEO (Generative Engine Optimization). Brands no longer compete to appear on Google but to be the only answer the AI offers to the user. It’s a logical but concerning evolution. If companies can pay to have their products be the preferred response, the line between information and advertising blurs. The "dark patterns" in conversation add complexity. Unlike visual tricks on websites, these are psychological and hard to detect. The chatbot can suggest products empathetically: "I understand you’re tired; maybe this supplement will help..." Artificial intimacy becomes a sales tool. Moreover, sponsored links appear with one click, while organic sources require more steps. Experts call it "selective friction." August 2, 2026, will be crucial: the EU AI Act demands total transparency in recommendation algorithms and advertising. All AI-generated content for commercial purposes must be clearly marked, even in metadata, following standards like C2PA. Penalties are severe: up to 3% of global annual revenue or €15 million, whichever is greater. The law prohibits manipulative techniques that affect user decisions. If ChatGPT uses your emotional state to show you ads, it could be violating the law. The Code of Good Practices, published this January, requires companies like OpenAI to ensure total traceability: users must always know when they are interacting with AI and when with sponsored content. The question is whether oversight will be effective and fines deterrent. If AI knows users so well, advertising stops being a simple ad and becomes a highly persuasive suggestion, tailored to your needs and weaknesses. It’s not a poster; it’s a salesperson who has studied your profile for months. We will need to watch if the mass inclusion of advertising causes AI to sacrifice quality and intelligence for profitability, prioritizing paid sources over academic or open-source ones.