Ігор Вишневський AI Eng 19 March 2026, 09:00

Advertising in ChatGPT. Will the world's most popular AI chatbot start "feeding" users "commercial" answers? What consequences could this have? We analyzed together with AI experts

OpenAI has long denied the possibility of advertising in its ChatGPT. Moreover, Sam Altman previously stated that advertising is a «last resort» for his company, and it is the combination of advertising and artificial intelligence that worries him.

But not so long ago, Altman’s «concerns» were completely dispelled: the company not only acknowledged its intention to introduce ad blocks, but also launched test advertising in the US in the free version of the chatbot. Currently, such advertising is presented as separate content blocks with the inscriptions «sponsored».

ChatGPT’s «advertising» responses — conspiracy theory or real developments

Investor, entrepreneur, and AI expert Mykhailo Patsan does not consider such a potential turn of events to be exaggerated fears — in his opinion, it is an absolutely real risk.

«The advertising model by its nature creates a conflict of interest between response quality and monetization. We’ve already seen this scenario with search engines: Google started as a neutral search, and today the first page of results is mostly advertising and SEO-optimized content, not the best answer to the query. ChatGPT has all the data to make targeting much more accurate: it knows not just your queries, but your context, your problems, your level of decision-making. This is an advertiser’s dream,» Patsan says in a comment to our publication.

He admits that technically the «adjustment» could be very subtle — not in the form of overt advertising, but simply a ranking of recommendations, where «paid partners appear a little higher or are described in a little more detail.» «The user won’t even notice it,» says Mykhailo Patsan.

AI expert and online marketer Ivan Blazhnov is convinced that OpenAI, of course, has the technical ability to introduce a certain advertising context into the chatbot’s responses, but he doubts that it will do so.

«GhatGPT has data about the consumer, and the data it received from the advertiser. Comparing this data and taking it into account in a certain way when preparing a response — yes, it is possible. But will they do it? It is unlikely,» the expert doubts.

In his opinion, with such a move, OpenAI could violate a large number of restrictive regulations, and «bringing this to light» would be quite easy.

«This can be easily verified. It is enough to register several accounts, provide some information about yourself from them, and understand which advertisers are shown in each of these accounts. You can specifically ask such questions to understand whether he will advise this or that advertiser more often — it is quite easy to check with a small amount of money,» Ivan Blazhnov is convinced.

According to the expert, if OpenAI starts to covertly introduce advertising context into the chatbot’s responses, the company risks much more than the benefits it can receive from advertisers.

AI expert Alexey Minakov also admits that, purely theoretically, such «advertising-colored» correction of answers is possible, «since the processes of forming answers themselves are not open and transparent.» In this context, he gives a simple example of another popular AI service, Grok, which has been repeatedly caught «broadcasting the opinion» of its owner, Elon Musk, in its answers.

«Grok has repeatedly come across the fact that it takes into account the owner’s opinion when generating answers. But now it is about the fact that the answers will not be adjusted, and the advertising will be in a separate block. That is, it will not be „product placement“, natively or covertly integrated into the answers, it will be a separate block,» the expert says.

Not «hidden promotion,» but personalization of responses

DevRain CEO and AI expert, Microsoft AI MVP Oleksandr Krakovetskyi suggests approaching this discussion from a slightly different perspective.

«It all depends on how you view AI. If you perceive AI as a threat, then it makes sense to view advertising integrations with fear (evil corporations are telling me what to drink, eat, how to dress, and what movies to watch). If you perceive AI as the best technology in human history that can provide personalized answers, then people should be very interested in seeing advertising appear in answers as soon as possible,» he expresses an unconventional position.

As Krakovetsky says, AI models are built to default to the most popular (neutral) answer, based on top (trusted) sources. Therefore, most answers, he says, contain general advice, well-known services, or approaches.

«Thus, you are unlikely to learn about alternatives, new products, services, etc. „out of the box“ without additional research. What movie to watch? Top IMDB. What video conferencing tool? Zoom. What drink? Coca-Cola or Pepsi. What is the best series? Breaking Bad. No personalization! Advertising integrations can change this. Here are popular services, and here are alternative, albeit advertising ones. Here are famous films, because they are high on the IMDB list, and here try this one — you will tell me later whether you liked it. After many recommendations, AI services will clearly understand what to offer,» says Krakovetsky.

He gave an example from his own work practice: it was necessary to implement some automation, but there was no functionality «out of the box».

«ChatGPT offered various non-working or suboptimal options, and only after I clearly said „let’s not just use built-in options“ did they move on to recommending paid alternatives. If the answer was „it’s impossible to do it out of the box, here are some paid options“, it would save everyone nerves and time,» CEO DevRain is convinced.

Krakovetsky does not deny the possibility of abuse in this process, but reminds us that it is now practically impossible to get to the first page of Google without huge budgets.

«Of course, I don’t know how advertising will work in a year, two, or three in generative services, but I’m impressed by the potential,» the expert suggests looking at it from a positive perspective.

Is it still a step towards the degradation of AI?

However, not everyone agrees with Krakovetsky’s position. Mykhailo Patsan believes that if this «advertising» logic wins, it will lead not just to degradation, but to a fundamental change in the nature of the AI tool.

«ChatGPT is valuable today precisely because it has no incentive to lie to you for profit. It does not sell you insurance, does not take you to a certain store, does not lobby for someone’s interests. This is a unique position that no media, no consultant, no search engine can offer today. As soon as an advertiser appears, a third party with its own interests appears inside your dialogue. And you no longer know where an honest answer ends and commercial interest begins. This destroys trust — the only asset on which the entire value of such tools is built,» Mykhailo Patsan is convinced.

Thus, in his opinion, short-term advertising revenues can turn out to be «very expensive in the long run.»

AI expert and online marketer Ivan Blazhnov claims that the very idea that chatbot responses can include promotional information and advertisements is quite predictable, because ChatGPT is already «biting off» a fairly significant number of Google search queries. On the other hand, with the advent of contextual advertising in Google, search itself did not degrade. So this should not happen to ChatGPT either.

«Advertisers are used to the fact that using the contextual advertising tool, you can directly show certain ads to those who are currently searching for this or that information, relying on keywords. Since it is these users who use search that are now following this in ChatGPT, it would be quite logical to provide users with the advertising that they are used to receiving in Google search, advertisers with a tool for advertising, and themselves to earn additional money from it. But the contextual advertising that appeared in the search results — I would not say that it led to its degradation,» Blazhnov concludes.

So, if ChatGPT or other neurons appear with something similar to contextual advertising, then, according to the expert, it is unlikely to lead to such «distortions» that people stop using these tools, and the neurons themselves stop providing users with specific useful answers.

«This makes no sense — especially to the companies that founded neural networks,» he believes.

Safeguards against «advertising degradation»

AI expert Alexey Minakov reminds us that OpenAI’s competitors, in particular Anthropic, the developers of Claude, have already clearly stated that their service will be ad-free. «That is, they will use other monetization models — from subscriptions to tariffs to payments for using their models via API,» he says. However, OpenAI has previously stated that it will be ad-free, which did not prevent it from changing its mind at some point.

According to Mykhailo Patsan, to avoid the risks discussed in the article, OpenAI and all other AI companies should take several specific steps, «without which any declarations will remain empty words.»

The first thing that needs to be implemented is a clear architectural separation between the ad unit and response generation, according to Patsan.

«Advertising should be a separate layer of the interface, not input data for the model. Technically, it is implemented — it is only a matter of business will,» he believes.

The second mandatory step, according to Mykhailo Patsan, is a public audit and independent verification of responses for bias.

«Just as pharmaceutical companies are required to publish clinical studies, AI companies should provide access to independent researchers to check for systematic biases in responses,» the expert is convinced.

The third step to protect against advertising manipulation in AI chatbot responses, according to Mykhailo Patsan, is a transparent monetization model for the user.

«If a specific brand or product is mentioned in a response as part of an advertising agreement, it should be marked as clearly as „advertisement“ in regular advertising,» he emphasizes.

And the last thing that Mykhailo Patsan emphasizes in this context is a clear regulatory standard for the industry. «Self-regulation in this area does not work. Mandatory disclosure standards are needed, similar to those that exist in financial advertising or media. As long as there are no such standards, each company will pull the blanket towards monetization, justifying it by «improving the user experience,» the expert emphasizes.

Ivan Blazhnov, meanwhile, believes that the best safeguard against manipulation is high competition in the field of AI, which simply will not allow any of the big players to risk their reputation.

«Since there are not one or two players in this market, and the competition is quite significant, this is the best safeguard against one single strong player introducing certain distortions into the answers it provides. This will be immediately noticeable in comparison with other models, and people who are used to using neural networks will immediately feel it. If the changes are significant, people will switch from one provider of „neural chats“ to another,» he says.

Thus, the AI expert concludes that it is precisely due to high competition that the risks of AI degradation under the influence of advertisers are not very realistic.

«Due to the fact that there are several players in the market, and they are all strong, and they are all fighting to increase their user base, this is the best barrier to any of these players using the existing user base with certain distortions in the effectiveness of responses in favor of situational benefits from advertisers,» Blazhnov concluded.

ChatGPT can now create interactive visual explanations for learning math and science

An American entrusted ChatGPT with the sale of his house: how AI coped with the task

Checking chatbots for fakes: a comparison of ChatGPT Gemini and Claude on the same prompts showed who most often makes up details in the news