Imagine everything you shared with ChatGPT—whether relating to your health, fears, vulnerabilities, or intimacies—is, in fact, not private. Imagine instead that it is now material for an exceptionally precise advertising system, one that analyses you intelligently and in the most personal way. With ChatGPT creator OpenAI moving towards advertising, there is real concern that queries asked on trust could feed marketing channels that draw on users’ most sensitive data.
Since its launch in late 2022, ChatGPT has been presented as an alternative to the traditional internet, with no promotional links and no stream of adverts, but a direct knowledge assistant that keeps data confidential. The absence of advertising was central to its identity and key to building trust with the hundreds of millions who turned to it for advice and, often, for a better understanding of themselves.
On 16 January, the company said it had begun testing advertising in the United States within the free version and a lower-cost subscription tier. It said adverts would be clearly separated from answers, would not appear alongside sensitive topics, and would provide advertisers with broad indicators rather than individual data. Its decision marks a radical shift, opening the door to an advertising model inside an interactive AI environment for the first time.
Watershed moment
The change goes beyond a single company and is a defining moment in the evolution of the internet. Just as search engines reshaped the digital economy around advertising, AI systems may reshape the digital economy around the answer itself. How AI will generate profits is now better understood, but what remains to be seen is how that will affect the nature of knowledge and trust, and the user who may find that the most private aspects of their life are now grist to the advertising mill.
A move to introduce advertising in ChatGPT can be understood by looking at the large gap between revenue and expenditure in the AI sector. Although OpenAI is valued at around $500bn and generated around $20bn in revenue in 2025, these figures are dwarfed by its financial commitments: roughly $1.4tn on computing infrastructure over the next ten years, with cumulative losses expected to reach $143bn before the first profits are projected in 2029.
These indicators suggest that, despite ChatGPT having more than 800 million weekly users, the AI model has not yet achieved financial self-sufficiency, as it still needs additional income. This model differs fundamentally from traditional internet companies. While social media platforms relied on user-generated content at limited cost, every interaction with an AI system carries direct energy, processing, and infrastructure costs. Higher user numbers drive costs up, rather than down.
In this context, advertising appears to be economic logic, even if it represents a departure from the platform’s philosophy (OpenAI boss Sam Altman previously described it as a last resort). Estimates point to annual advertising revenue of $25bn by 2030. Yet the decision also reflects a broader reality. AI has become a costly global infrastructure project, and unlike companies like Google, which fund technology development through established advertising profits, OpenAI does not have a comparable profit base.
