A shopping prompt inside ChatGPT has raised a simple but difficult question for OpenAI: when does a helpful product suggestion become an ad?
The issue surfaced after paid ChatGPT users recently reported seeing a message labeled "Shop for home and groceries. Connect Target." The prompt appeared to invite users to connect their ChatGPT account to OpenAI's retail partner Target. To many of the people who saw it, the experience looked commercial, even if OpenAI does not describe it that way.
What users saw
The reported prompt was short, direct, and tied to a named retail partner. It did not appear as a generic product research feature. It appeared to point users toward Target, a company identified in the source as OpenAI's retail partner.
That detail is why the reaction matters. A shopping suggestion in a conversational assistant can feel different from a tool that simply helps compare products. When a named partner appears in the interface, users may reasonably wonder whether the assistant is recommending a useful action, promoting a commercial relationship, or testing a new way to make money.
OpenAI pushed back on that interpretation. ChatGPT Product Manager Nick Turley said there are "no live tests for ads" and argued that screenshots circulating online "are either not real or not ads." His comments reject the idea that OpenAI is currently running advertising tests inside ChatGPT.
At the same time, the source notes that his wording indirectly points to the possibility of internal tests. That distinction is important because OpenAI is not only disputing what users think they saw. It is also drawing a boundary around what the company considers an ad.
OpenAI's internal line is not obvious to users
Chief researcher Mark Chen took a more careful position. He acknowledged that the Target prompt could feel like an ad and said these displays need careful handling. According to Chen, OpenAI has already disabled the feature.
Chen also said OpenAI is working on improving the model's precision and on settings that would let users reduce or completely turn off similar suggestions. That response suggests the company understands the problem is not only technical. It is also about user control and trust.
The gap between OpenAI's definition and a user's experience is the center of the dispute. A company may see a prompt as a feature, integration, suggestion, or experiment. A user may see a named retailer and read the same interface as advertising.
That gap becomes larger inside ChatGPT because the assistant is not a typical page of links. The source notes that many people use ChatGPT as a reliable and even personal adviser. In that setting, a product recommendation can carry more weight than a sponsored link on a search results page.
The business pressure behind shopping prompts
The debate is also happening against a broader financial backdrop. The source states that roughly 95 percent of ChatGPT users are on the free tier. That creates pressure for OpenAI to find ways to monetize the product without alienating the people who use it.
Shopping prompts would be an obvious route toward revenue because they connect user intent with possible purchases. But they also create risk. If users begin to suspect that suggestions are shaped by commercial partnerships, they may question whether the assistant is optimizing for their needs or for OpenAI's business model.
This is why OpenAI's wording matters. Saying something is not an ad may be technically meaningful inside the company, but it may not resolve the concern for users. A prompt can feel promotional even if it is not sold, priced, labeled, or measured as a traditional advertisement.
The source also points to a past warning from CEO Sam Altman. He described a future in which ChatGPT might tell users "You should think about buying this product" or "You should go here for vacation" as dystopian. He also emphasized a preference for a simple business model where paying users were not the product.
That earlier position now frames the current debate. The reported Target prompt did not appear in isolation. It arrived as OpenAI is developing shopping-related capabilities and facing the challenge of monetizing a massive product.
Memory makes the question sharper
The source says OpenAI's new Shopping Research agent already uses ChatGPT's memory system to personalize product recommendations. It also says internal discussions reported by The Information suggest the company has explored using memory for targeted advertising.
Those facts make the line between recommendation and advertising even more sensitive. A general shopping assistant can help users compare options. A personalized shopping assistant can use remembered preferences to shape what appears next.
Personalization is not automatically advertising. But when personalization, product recommendations, and commercial partners appear in the same product, users need clear explanations. They need to understand why a suggestion appeared, whether a partner relationship influenced it, and how to reduce or turn off similar prompts.
Chen's comments point in that direction by mentioning precision improvements and user settings. But the larger issue remains unresolved. OpenAI will likely need a clearer public definition of what it counts as advertising, especially if ChatGPT continues to blend product research with commercial integrations.
Why the Target prompt matters
The Target prompt is important because it tests more than interface design. It tests how much ambiguity users will accept inside an AI assistant they may treat as a guide, planner, researcher, or adviser.
OpenAI says the reported screenshots are not evidence of live ad testing. Some users saw a shopping prompt tied to Target and reached a different conclusion. Both facts can exist at the same time, and that is exactly why the issue is difficult.
For OpenAI, the practical challenge is not only whether a prompt meets an internal definition of advertising. It is whether users can tell when ChatGPT is helping, when it is recommending, and when a business relationship may be part of the experience.
Until that distinction is clearer, even a single shopping suggestion can become a larger signal. It shows the pressure OpenAI faces as it tries to turn ChatGPT into a broader commercial platform while maintaining confidence in the assistant's advice.