On September 15, at the "Cyber Security Awareness and Literacy Improvement" sub-forum of the 2026 National Cyber Security Publicity Week, Douyin announced that it will promote the "Trusted Community Guardian Plan" and the "Friendly Community Co-construction Plan" based on the platform's systematic governance, and provide users with a more reliable information environment and a friendlier communication space through the improvement of technical capabilities, operational guidance and rule construction. "Trusted Community" aims to solve content authenticity issues such as rumors, false staged photos, AIGC abuse and malicious marketing; "Friendly Community" focuses on online violence, anger baiting and the management of polarized views. The above two plans together form Douyin’s “trustworthy and friendly” community building framework.
In the AI era, network ecological governance faces unprecedented challenges. It may only take a few seconds for a rumor to be generated. In addition to fake photos to "disguise the truth", it can also be copied in batches. Confrontational rhetoric, online violent comments, anger bait, etc., naturally adapt to attention, and often make people unable to distinguish between true and false, and call them cheated. Under "technological equality", the threshold for content production has been lowered repeatedly, and there are more and more outlets for dissemination. A fake video often breaks across platforms, and a bad comment can usually set off a storm that overwhelms ordinary people. Many Internet platforms are troubled by "breaking rumors" while at the same time being "overcrowded with reports"; users are gradually diluting mutual trust in half-truths and half-false information. When falsehoods and malice enter the stage of "industrialization," low-cost, and large-scale, if platform governance is no longer satisfied with measures such as single-point account bans, subsequent deletions, and current restrictions, it will obviously be unable to keep up with changes.
In this context, Douyin splits the governance and operation of online communities into two fulcrums: "credibility" and "friendliness". The idea is very clear: the former focuses on whether the content is true, and the latter focuses on whether the communication is smooth and kind. What they both target together are rumors, false staged photos, AIGC abuse, malicious marketing, online violence, anger bait and other stubborn Internet diseases. It is worth mentioning that this time the platform focuses on "defeating magic with magic" through upgrading AI model technology and refined rules construction. If you use AI to spread rumors, I will use models to identify them; if you use so-called tools to spread rumors, I will use data and algorithms to verify and uncover them. In response to rumors, staged photos and other situations, Douyin has improved the "truth-seeking" large model product function, superimposed product functions and AI technology empowerment, so that the content must pass the "fact verification" before running. In response to malicious provocations, anger bait, polarized views, etc., Douyin has introduced two scoring dimensions: "positive review target" and "hate review target" in the recommendation system. Content such as provocations and negative reviews will not only receive no distribution revenue, but will also be restricted in traffic and reduced in rights. In other words, governance is not about waiting for reports and complaints to come before taking action, but about making false content and emotional garbage unable to be counted and pushed out within a scientifically operated mechanism.
From technology identification to rule design, from content distribution to community atmosphere construction, platform governance has entered the stage of "systematic operations". However, a series of practical experiences have shown that the real value of AI governance lies not in “deleting more and shutting down faster”, but in allowing risks to be discovered earlier and the facts to reach users earlier. Of course, we must also clearly see that no matter how powerful the platform technology and rules are, there are still objective problems that are beyond the reach of others: cross-platform transfers lengthen the chain of responsibility, iteration of black production techniques is faster than rule construction, ordinary users cannot distinguish between "AI synthesis" and "live direct shooting", and creators are constantly swinging between traffic and the bottom line. Therefore, it is only the "first half" for the platform to provide technology and establish rules. The longer and more effective "second half" still needs to focus on multi-party governance. Users should not be "angry forwarders" and wait and check when disputes arise; creators should not use anger as a persona and staged photos as "passwords"; no matter how good the traffic is, they must not indulge in "fake rice"; industry partners and institutions should be more rational and self-disciplined, and strictly restrict the behavior of individuals and institutions; regulatory authorities should comply with laws and regulations, insist on cracking down on the joint liability of relevant institutions and accounts, and malicious marketing of gray and black products, and investigate to the end.
The Internet has never been a lawless place. Credibility means respecting facts; friendliness means protecting the rights and interests of users and the network environment. When more platforms are willing to take more responsibility, and when more users are willing to pause for three seconds before forwarding, the openness and excitement of the online world will be further stimulated. This kind of online space benefits not only the platform, but also every user in front of and behind the screen.
