%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
On May 25, the Douyin platform released its annual report on the governance of rumors, revealing the latest achievements of AI large model technology in the information content ecosystem. Since introducing the "AI Quzhen Large Model" a year ago, Douyin has made a significant breakthrough in combating false information, shifting from "people finding rumors" to "models taking proactive action."
Data shows that in the past year, the average views of rumors identified and handled by the AI Quzhen Large Model have dropped significantly by 62%. This means the "window period" for rumors to spread before being effectively intercepted has been greatly shortened, cutting off the information dissemination chain at an early stage.
Technology Empowerment: From Semantic Verification to Proactive Defense
According to the introduction, the "AI Quzhen Large Model" on Douyin now has DeepResearch (deep research) and multi-dimensional cross-verification capabilities. When handling homogenized rumors with multiple "variants," such as the rumor about "monkeys shot dead on Emei Mountain," the model can automatically associate and handle them through semantic recognition, greatly reducing the workload of reviewers.
The core technical highlights include:
Proactive Governance: The accuracy of rumor identification has increased to 95%, and the processing timeliness for hot events this year has improved by more than 50% compared to before.
Dual Protection: While retrieving authoritative sources, the large model deeply integrates Douyin's years of accumulated manual review experience, ensuring accurate judgment in complex logical analysis.
Cognitive Defense: Through the "Quzhen Card" feature, the platform directly presents verified information from authoritative institutions and mainstream media in the form of visual cards below controversial videos or in search results, guiding users to access factual truths.
Value Return: Multi-party Collaboration to Build a Co-governance Ecosystem
A person in charge of Douyin's misinformation governance stated that the introduction of AI technology aims to build a "cognitive defense" for users, but the ultimate foundation of rumor governance still lies in news authenticity. The effectiveness of the large model's judgment depends on timely and accurate information clarification from news subjects, authoritative official agencies, and media.
In this report, Douyin also reviewed the top ten debunked cases over the past year. These cases covered public safety, health science, and social hot topics, each of which was traced back and verified through AI联动 with authoritative sources. Currently, when users browse controversial content or search related keywords, the system will automatically trigger the "AI Quzhen" prompt, providing a complete fact溯源 link.
Industry experts said that Douyin's upgrade in its governance strategy not only demonstrates the engineering application of large models in the field of content security but also marks that short video platforms are transitioning from simple "content blocking" to a "collaborative governance" model that cultivates users' discernment and builds a clean community together.