Abstract: For emerging technologies that have the ability of self-learning, self-thinking and self-decision- making such as generative artificial intelligence, anti-monopoly should shift from monopoly behavior regulation to monopoly risk governance. Because of the dual technical nature of both consuming products and technical models, generative artificial intelligence can not only monopolize data, build competition barriers and exploit consumers with its independent consciousness, but also continue to enable tech giants as controllers of generative artificial intelligence to consolidate monopoly status and implement monopoly behaviors. It has predictable monopoly risk on the dimensions of technology and application. In this regard, anti-monopoly should adhere to the scenario-based management of two-dimensional monopoly risk on the basis of maintaining and improving competitiveness of the market. For generative artificial intelligence itself, through learning from the EU Artificial Intelligence Act, monopoly risk can be categorized and classified according to the operational mechanism of generative artificial intelligence. For generative artificial intelligence controllers, targeted scientific governance paths should be built in the light of prior compliance governance, in-process governance and post-regulation. In general, it is necessary to strengthen the anti-monopoly penetrated regulation on the technology and the entity.

Key words: generative artificial intelligence, monopoly risk, category- and class- based governance, penetrated regulation