According to a report from the Cyberspace Administration of China, in June 2025, online rumors will mainly focus on social hotspots, disaster and accident situations, pseudo science popularization, and other fields. Rumormongers will fabricate or exaggerate facts through means such as cross dressing, AI generation, and pseudo scientific packaging, disrupting normal social order. The report specifically mentioned the destructive power of AI generated false information in disaster and accident rumors. Some netizens took the opportunity to fabricate false videos, pictures and texts such as "the extremely heavy rainstorm in Lanshan County, Yongzhou, Hunan caused the car to be washed away", "serious water accumulation on the road surface in Jianoufangdao Town, Nanping, Fujian", and "sudden mountain torrents in Banfanggou, Nanshan, Urumqi". Such rumors not only exacerbate social panic, but also disrupt normal disaster prevention and relief order, causing endless harm. AI generation is widely present in various fields such as information dissemination and content creation. While it brings convenience to work and life, enriches creative methods, and expands reading space, it also poses great challenges to information dissemination. The deep synthesis videos of forged celebrity speeches, "fake science popularization" articles that deceive the real, as well as various photos and videos of the scene of the incident, are eroding the public's ability to discern information through AI based fake information with its high degree of simulation. It can be said that the problem of information fraud has become a cancer of the Internet, and failure to govern will erode the trust foundation of the network society. No matter from which perspective, it requires us to strictly distinguish AI content and let AI synthesized content return to its original appearance. Faced with the new problem of AI synthesis fraud, it is necessary to upgrade management methods. The commonly used prevention and control method at present is to label and make AI synthesized content public and exposed to sunlight. Currently, the reason why AI can create fake information and cause trouble is largely because it disguises itself well, making it difficult for others to see through and exploit loopholes between truth and falsehood. Once the AI synthesized content is clearly labeled, this space naturally disappears. The relevant laws and regulations are being improved. The "Artificial Intelligence Generated Composite Content Identification Measures" formulated and implemented on September 1st this year further provide detailed regulations on how to identify. At that time, relevant laws and regulations will form a combination to plug the loopholes in the governance of AI synthesized content. At the same time, the platform should further supervise and constrain authors to strictly comply with relevant regulations, establish a regulatory mechanism that is traceable in terms of identification, technically reversible, and clearly defined in terms of legal responsibility, so that the platform's algorithm crawling and recommendation can become a model of good governance in cyberspace, rather than an accomplice to the spread of rumors. In addition, AI model companies should actively participate in governance, and it is necessary for model companies to embed tamper proof identification modules at the technical level. Historical experience shows that every technological revolution is accompanied by an iteration of governance models. From intellectual property protection in the era of print to cyberspace governance in the era of Internet, mankind has always been looking for a balance between innovation and order. The foundation of the Internet society is honesty. Only by keeping the bottom line of true communication, can the Internet maintain its vitality in technological innovation. (New Society)