AI (Artificial Intelligence) has learned to advertise. Imagine a scenario where you want to give a friend a birthday gift, but are unsure of what to choose, so you seek advice from AI. AI retrieves a large amount of data to provide answers, but upon closer inspection, the information sources on which recommendations are based are often similar to advertising soft articles. At this moment, one cannot help but wonder - why does this AI seem like a 'salesperson'? This scene is no longer imagined, but a reality. Recently, media surveys have found that GEO (Generative Engine Optimization) services are increasingly favored by businesses. This type of service analyzes the content crawling preferences of AI models and uses a set of technical settings to integrate content with brand information into answers. Simply put, it is the artificial control of AI's information filtering. In the advertising industry, this model is not new. For example, in the era where search engines dominate traffic entry, there are services that optimize website content based on search engine ranking rules. At present, user preferences are gathering towards AI, and AI generated content will inevitably become a focus of business competition. Reasonably utilizing rules to engage in competition is in line with business logic. But the problem is that compared to common advertising channels in the past, using GEO services to let AI "advertise" has generated more compliance controversies. The "selling point" of GEO services is largely "deceptive", which means that commercial promotion is "packaged" into AI generated objective content to enhance advertising persuasiveness. However, the Advertising Law clearly stipulates that "advertisements should have recognizability that enables consumers to identify them as advertisements". AI advertising that is difficult to identify may infringe upon consumers' right to know. Some GEO service providers fabricate facts to attract the attention of AI, and are also suspected of false advertising and unfair competition. In addition, the development of AI big models cannot be separated from the cornerstone of high-quality training data. If a large amount of homogenized information mixed with commercial attributes is allowed to pollute the "source water" of AI training, it will not be conducive to the iterative evolution of technology in the long run. The Central Economic Work Conference to be held at the end of 2025 proposed to improve the governance of artificial intelligence. How to solve the problem? It should be recognized that in the face of new technologies, new scenarios, and new problems, relying solely on old experiences and methods is not feasible. For example, from the perspective of technological evolution, the application of AI means that it will replace humans in making many specific decisions. The recently discussed "AI mobile phone" replaces users with intelligent agents for ordering, shopping, etc., and users only need to confirm payment. When technology becomes a 'black box', users are more likely to be 'led by the nose' unconsciously. For another example, the traditional Internet advertising supervision is mostly post supervision, and the mechanism of supervision and prevention in the "feeding" stage of AI data needs to be improved. Therefore, for those in charge of governance, the urgent task is to keep up with the times, transform their governance thinking, promote the advancement of checkpoints, and improve the efficiency of the entire process supervision. Assess the characteristics of new business models and establish clear red lines; Clarify the responsibilities of service providers and platforms, and eliminate "gray areas"; Exploring the use of technology to govern technology, improving the transparency and interpretability of related algorithms... Only by being flexible and well managed can we balance the application of technology with the protection of user rights. From a broader perspective, AI "advertising" reflects the common challenges faced by governance in the wave of technological change. The advanced nature of technological development tests the ability of governance to respond quickly and follow up in a timely manner; The professionalization of technical issues calls for a more open and diverse governance structure to absorb more professional wisdom for shared governance. But ultimately, what remains unchanged is the courage to reform and the open mindedness. These are the experiences we have gained from recognizing, adapting to, and seeking change in the past, as well as the confidence we will have to take the initiative in the future. The technological vision does not always translate into a beautiful reality. People need to constantly reflect on and correct the direction of technology in the confrontation of viewpoints and the game of interests, avoid falling into the narrowness and arrogance of technologism, and enable the whole society to share prosperity. AI 'advertising' may be one aspect of this complex game. Only by pursuing progress, embracing innovation but never letting it go unchecked, and running blindly, can we ensure that technology brings universal benefits with human subjectivity. (New Society)