The responsibilities and challenges faced by the Douyin platform in managing online harm

The importance of addressing online harm on the Douyin platform:

In recent years, the issue of online harm, ranging from the controversial “human flesh search” to hate speech and discrimination against specific groups on social media, appears unending. In December 2021, the search for biological parents by 15-year-old Liu Xuezhou on the Douyin platform garnered significant attention. But when Liu Xuezhou used the Internet to find his parents, disaster had just begun to befall the boy. Following his parents’ slanderous online remarks about Liu Xuezhou, public hate speech against him ensued. Since 2022, several netizens have questioned that Liu Xuezhou’s purpose of searching for his biological parents is not simple, but for money. Liu Xuezhou started a live broadcast on Douyin to clarify the matter, but many people who made negative comments entered the live broadcast room. Liu Xuezhou has suffered abuse from the Internet for a long time.

In the early morning of January 24, 2022, Liu Xuezhou committed suicide by taking medicine in Sanya, Hainan, and eventually died after rescue efforts failed. After the incident ended with Liu Xuezhou’s suicide, people were extremely angry at the online abusers and hoped that the platform and the law could punish these online abusers. Liu Xuezhou’s aunt sued two Douyin bloggers, who have millions of followers. Before Liu Xuezhou committed suicide, these two bloggers published offensive language and malicious speculation many times, which caused great harm to Liu Xuezhou.

The Liu Xuezhou incident is just a microcosm of countless online harm incidents. According to Flew (2021), the escalation of online hate speech via digital platforms and social media is increasingly recognized as a significant issue. In this incident, Liu Xuezhou also used other social media such as Weibo, but Douyin was Liu Xuezhou’s main social media platform. The main reason is that douyin, as a short video-based platform, will have 750 million users in 2023 (Statista, 2023). Liu Xuezhou’s original wish was to find his parents, so he hoped to use the huge exposure of the douyin platform to let more people see the news about finding his parents. However, this incident underscores how the vast exposure of social platforms can be a double-edged sword. On the one hand, Douyin’s huge influence can help Liu Xuezhou find his parents, but when it is used maliciously, large-scale hate speech will occur. Flew (2021) believes that it is crucial whether platform companies can moderate their content in the public interest and whether they demonstrate government involvement.

Challenges Douyin faces in dealing with online harm:

Online hate speech represents an escalating issue within the contemporary digital society. Li Yan, Douyin’s ecological product manager, talked about some of the difficulties Douyin faces in managing hate speech.

A key characteristic of hate speech is the rapid escalation of intense emotions within a brief timeframe. It is difficult to quickly and accurately distinguish between hate speech and non-hate speech. Although platforms try to “clearly” define harassment and/or hate speech on their platforms, harassment and hate speech are often the most controversial forms of harmful content (Pohjonen, 2019). There are four reasons why it is difficult to distinguish between hate speech:

1. Dialogue roles influence the perceived intensity of language violence. A sentence deemed humorous among family, friends, and partners can appear hostile to strangers.

2. Perceptions of online language vary significantly across individuals, making the boundary between hate speech and non-hate speech nebulous. It hinges on the subjective experiences of the individual, complicating universal definition and identification.

3. It is difficult to quantify the speed and quantity of hate speech messages. For example, after a user receives hate speech, when should the platform protect the user? Is it when the user receives the first piece of hate speech? Or when a user receives their 500th piece of hate speech? Douyin believes that developing a clear digital standard is difficult at this stage.

4. Public institutions and public figures are bound to bear more criticism. Because public figures are concerned by all parties and have higher requirements, they also bear more responsibilities. The definition of online harm for this group is controversial.

Secondly, Douyin’s daily active users are huge. Therefore, the review workload faced by Douyin staff is huge. Errors or omissions in moderation can swiftly result in public discontent. At the same time, even relying on algorithms to review hate speech requires a lot of effort because all these models rely on human-annotated training data, which is a laborious process and contains potential biases. Waseem (2016) found that amateur annotators were more inclined than their trained counterparts to classify messages as hate speech.

1. When major events occur, a large number of online comments will appear in a short period of time. For example, within 12 hours of the Tangshan incident in China, Douyin received nearly 3 million comments. Even if the machine initially reviews 2 million comments, the remaining 1 million comments will require 400 reviewers to work continuously and intensively for 12 hours to complete.

2. The content of network harm is difficult for machines to identify. Use homophones and abbreviations to post provocative and sarcastic remarks or edited pictures. These contents fall somewhere between criticism and online harm, and are difficult for machines to identify.

3. Verbal violence is subject to varying degrees of punitive measures. For example, when users publish content that contains curses, attacks, exposures, ridicules, and provocations, reviewers need to punish them accordingly. However, whether to ban an account indefinitely, ban it for a short period of time, or simply delete comments all need to be carefully evaluated according to the degree of verbal violence.

4. The authenticity of external information is constantly changing. This also has an impact on the processing of online harm when events are reversed.

Third, managing online harm is unique compared with other Internet governance scenarios. This will be further elaborated on in three points.

1. After a large-scale emotional outburst, rational thinking can easily cause secondary online harms. High-profile instances have seen online harassment escalate into widespread hate speech. (Flew, 2021) In many cases of online harm, some famous commentators and Internet celebrities will express their opinions on hot events. Should their views not align with the majority, ridicule and subsequent online harm ensue.

2. Online harm is easily transferred across platforms. For instance, harm originating on platform A quickly proliferates to platforms B, C, and D. At present, few online harm incidents occur independently on a certain platform. For example, after the Liu Xuezhou incident, Weibo also publicly stated that they had suspended the private messaging function for more than 1,000 user accounts. These users sent private messages to Liu Xuezhou from January 1 to 24, 2022.

3. The process of managing network harm will involve public data and private data. Online harm scenarios often occur on private message pages, and the platform cannot actively monitor and analyze users’ private information.

To sum up, Douyin still faces many difficulties and challenges when dealing with online harm. Furthermore, balancing the principles of free speech with minimal censorship, alongside implementing effective legal measures against hate speech and online abuse, remains a critical challenge (Flew, 2021).

Douyin’s strategy for dealing with online harm:

Despite facing significant challenges in managing hate speech, Douyin has devised numerous strategies to mitigate online harm. These strategies are mainly developed from the perspectives of perpetrators and victims of hate speech.

First of all, Douyin’s main strategy against perpetrators is to intercept inappropriate comments through technical means. Additionally, perpetrators risk platform-imposed penalties after posting hate speech.

1. In order to deal with hate speech more accurately and efficiently, Douyin uses multiple scene features such as comments and contributions to identify and classify potential “violators”. For users with hate speech, the platform will remind users to speak rationally through private messages, pop-up windows, smart voice calls, etc. Douyin will ban users who continue to publish hateful speeches after being reminded, and the user will be publicized on the Internet. Over the past year, Douyin warned and handled 5,004 related incidents.

2. Temporary conversation setting function for private messages. When a user sends a message to a stranger, only one message can be sent to a single user at a time, and the stranger can also have a conversation with the user after the user replies. If the stranger does not reply to the user, the user can only send a second message after 24 hours. Such a function can try its best to block inappropriate comments.

Secondly, Douyin has developed a series of protection functions for victims of online harm, and these protection functions are still being upgraded.

1. Douyin has launched the “one-click protection” function. “Douyin Xiaoan” is Douyin’s safe smart assistant. When users are harmed by hate speech, users can send a message to “Douyin Xiaoan” to enable the “one-click protection” function. The “one-click protection” function includes “strangers cannot comment on my videos”, “strangers cannot send me private messages”, and “strangers cannot search me”. These functions are all settings related to protecting the privacy of personal accounts.

2. Douyin has launched functions such as comments, private message rights settings, and keyword blocking in the comment area for users. Users are afforded the capability to adjust the privacy settings for interactions within the comment section to one of three options: open to all users, limited to mutual followers, or completely disabled. Concurrently, Douyin proactively dispatches notifications to users upon the reception of private messages that may harbor potential risks.

3. When users face online harm, they may need to report a large number of private messages, so Douyin has also launched the function of reporting malicious messages in batches. At the same time, Douyin’s customer service phone number and reporting email address will also accept and handle reports of cyber violence in real time.

4. Victims who have experienced cyber violence need scientific psychological counseling. Douyin has established a psychological care team to provide psychological comfort. Douyin’s safety center has joined forces with ByteDance Charity, AI-lab and other teams to provide users with psychological care services.

5. Beginning in 2023, Douyin partnered with the China Legal Consultation Center to offer a legal consultation service targeting victims of hate speech. This service has helped a total of 208 users to solve hate speech and other related issues that occurred on Douyin. Because the public tends to panic when faced with large-scale hate speech. With legal consultation services, victims can use the law to protect their rights and interests.

These strategies reflect Douyin’s technological capabilities and sense of responsibility in controlling online violence. And the current protection function can prevent and control the occurrence of online violence to a certain extent. However, controlling online violence is not a task that can be completed in a short time, and it is definitely not a problem that the platform can solve unilaterally. Therefore, Douyin still needs to continuously invest a lot of energy to upgrade and improve the platform’s own ability to manage online violence.

Reduce hate speech through user education

On July 27, 2023, Douyin initiated the “Anti-Cyberbullying Campaign”. The platform garnered support from numerous celebrities and influential personalities across diverse sectors to enhance the campaign’s prominence. Participation in this initiative is open to all users. Upon joining, an indicator signifying their participation will be automatically visible to others.

Douyin has established accounts such as “Douyin Rules Mini Classroom” and “Douyin Rumor Refutation” to enhance users’ awareness of hate speech and improve their ability to protect themselves through the performance of some sitcoms, pictures and texts on the platform. The contents of these accounts take advantage of Douyin’s characteristics as a short video platform, using a simple and easy-to-understand method to educate users on “what is hate speech” and how to use the tools provided by the platform to protect themselves after being attacked by hate speech. At the same time, Douyin also clearly stated in the “Douyin Community Self-Discipline Agreement” that the platform prohibits and unwelcome user behavior.

At last

As a social platform with a large user base, Douyin has the responsibility and obligation to face the issue of hate speech. To avert tragedies akin to that of Liu Xuezhou, this blog employs Douyin as a case study to delve into the duties and obstacles platforms encounter in moderating hate speech. Over recent years, Douyin has refined its strategies to curb hate speech, yielding tangible outcomes. Nonetheless, Douyin, along with numerous social media platforms, continues to grapple with significant challenges in managing hate speech both “swiftly” and “precisely.” At the same time, managing online harm is not only the responsibility of the platform, but the government, judicial system, and the public also need to be involved. Douyin should also regard the management of online harm as a long-term investment and must be completed.


BBC. (2022, January 25). Liu Xuezhou: Outrage over death of “twice abandoned” China teen. BBC News.

Douyin. (2022, June 28). 18 paragraphs to understand how Douyin prevents and manages cyber violence. WeChat Official Account.

Douyin. (2022,September 13). Community guidelines. Retrieved April 8, 2024, from

Douyin. (2023, July 27). Today, Douyin launched an initiative that we hope everyone can join. WeChat Official Account.

Douyin. (2024, September 19). Douyin’s Anti-Internet Violence Legal Consulting Service Center has been online for three months and has provided services to 208 users. WeChat Official Account.

Douyin. (2024, October 1). Douyin Internet violence control this year: early warning and handling of 5004 related incidents. WeChat Official Account.

Flew,T. (2021).Hate speech and online abuse. In Regulating platforms (pp. 91-96). Cambridge, MA: Polity Press.

Gan, N., & Chang, W. (2022, September 23). Chinese man sentenced to 24 years for restaurant attack on women. CNN.

Media scroll. (2022, January 26). In the face of “Liu Xuezhou’s death”, suspending the private messaging function of more than 1,000 Weibo accounts is not enough.

Pohjonen, M. (2019). Extreme speech|A comparative approach to social media extreme speech: Online hate speech as media commentary. International Journal of Communication, 13(2019), 16.

Thomala, L. L. (2024, February 22). China: Douyin Maus 2023. Statista.

Waseem Z. (2016). “Are you a racist or am i seeing things? Annotator Influence on hate speech detection on twitter,” in Proceedings of 2016 EMNLP Workshop on Natural Language Processing and Computational Social Science (Austin, TX: Association for Computational Linguistics (ACL)), 138–142. 10.18653/v1/w16-5618

Be the first to comment

Leave a Reply