Generative Engine Optimization (GEO) has given rise to a novel form of data poisoning malpractice, which disrupts the order of fair competition. Compared with criminal sanctions, bringing such conduct under the regulation of the Anti-Unfair Competition Law of the People’s Republic of China better aligns with the digital economy’s developmental orientation of encouraging marketing paradigm innovation. Under the dual-intermediary competition structure of “operators—algorithms—consumers,” the legitimacy review of competitive conduct should shift from competitive outcomes to the means of intervention, thereby distinguishing legitimate commercial data injection from malicious data poisoning. In terms of object determination, the objectivity and traceability of information should serve as the foundation of legality for the inputs. In terms of conduct characterization, data poisoning aimed at securing one’s own recommendation weight, given its detachment from typical media and its lack of identifiability, fails to satisfy the prerequisites for the application of the Advertising Law and should therefore be regulated as false or misleading commercial publicity; data poisoning aimed at disparaging competitors should fall within the regulatory scope of commercial disparagement. Furthermore, large language model companies, as victims in their own right, should seek remedies under Article 1165 of the Civil Code, and they cannot invoke Article 13(2) of the Anti-Unfair Competition Law of the People’s Republic of China, which is designed to regulate technological sabotage. In response to the emerging trend of manipulating algorithms through cluster flooding with authentic information, Article 2 of the Anti-Unfair Competition Law of the People’s Republic of China should be activated to provide catch-all regulation, thereby maintaining long-term institutional resilience in the balance between encouraging marketing paradigm innovation and safeguarding the market competition order.