Advertising is increasingly likely to appear inside AI systems because providers need additional revenue to sustain their operations. A Bloomberg opinion piece warned about this risk last summer; less than a year later Anthropic spent $8 million in a Super Bowl ad—nearly 2.5 billion forints—to mock other AI companies for cluttering their systems with ads while positioning its own Claude as ad-free.
How big could the business be? Numbers and projections
Industry analyses estimate roughly $1 billion in ad spending on AI platforms in 2025, potentially rising 26-fold by 2029 and accounting for about 13.6% of the ads sold alongside online search. Since 2020, annual investment into the AI sector has been in the $200–300 billion range. By comparison, even the best-performing AI companies currently report revenues around $20–30 billion a year, and much of the sector remains heavily loss-making. Advertising offers a familiar escape route to profitability—Google now generates about $300 billion a year in ad revenue and Facebook about $200 billion—so AI providers are motivated to pursue ads as a revenue source.
Where ads have already appeared in AI
- Ads have appeared in ChatGPT in some countries (not yet in Hungary).\
- Microsoft’s Copilot includes ad-like elements; Microsoft claims they are 25% more effective than traditional online ads.\
- Google has started testing ads in Gemini.\
- Grok also contains ads; Elon Musk has said this is the future of X’s advertising business.\
- Chinese AI systems are moving in the same direction.
Anthropic, however, continues to position Claude as ad-free, a stance likely related to its strong subscription revenue performance.
Why AI looks attractive to advertisers
Generative AI can infer user intent from conversational context, enabling much more finely personalized recommendations than keyword-based systems. Ads that blend into a conversation may be perceived as helpful suggestions rather than overt promotions, increasing their effectiveness. But the very subtlety that makes these ads powerful also raises the risk of exploiting vulnerable groups—children, the elderly, people in crisis or with addictions—because such messaging may not be recognized as advertising.
The risk of enshittification
Experts warn about a lifecycle often described as enshittification: services start by focusing on users, then shift toward prioritizing business partners like advertisers, and finally optimize purely for profit at the expense of user experience. This pattern has been observed in social media; if AI platforms adopt similar strategies to chase ad revenue, user experience and trust could erode as platforms seek to maximize time spent and ad placements.
Decision-making and the "intention economy"
Researchers at the University of Cambridge note that people frequently ask AI for help in decision-making—what film to watch, what to cook, or what stance to take on public issues. Embedding paid ads into that context can influence choices; scholars call this the "intention economy." The study warns that AI can predict a user’s developing decision in real time and sell that information to interested companies, which could then steer the user’s choice with targeted advertising.
Regulatory and responsibility questions
Many questions remain unanswered: who is responsible if an ad on an AI platform misleads or harms users—the advertiser or the AI operator? How should political advertising be handled, and when does a message become political? How will systems indicate that a recommendation is paid content rather than an impartial response? Some large social platforms in Europe have banned political ads to avoid EU transparency rules, but that does not resolve the issue globally. More broadly, distinguishing between a genuinely autonomous AI suggestion and a paid placement will be challenging to verify.
Outlook: what to expect next
Ads in AI seem likely given the business pressures and massive capital flows into the sector. The conversational and highly personalized nature of AI recommendations introduces new ethical and regulatory challenges. Many experts fear these issues will be addressed only while platforms are already deploying ad models, or worse, only after harm occurs.
Summary
Bringing advertising into AI platforms is a predictable business move to address massive infrastructure and operating costs, but it combines strong commercial incentives with substantial risks. Conversational, personalized ads can be remarkably effective while also being more difficult for users to identify and resist, and they raise unresolved questions about transparency, responsibility, and protection for vulnerable groups. The crucial challenge is whether regulation and industry practices will keep pace with rapid deployment before serious problems accumulate.
