Author |Xiao Shijia
Editor |Qiao Qian
Ten years after Douyin exploded into the short-form video industry, long-form text, once considered inefficient and outdated, is once again gaining traction.
At the end of 2025, Douyin launched long-form text and image features, providing amateur creators and media organizations with detailed long-form creation entries and providing high-quality long-form text and image traffic support. Currently, users can only upload and publish articles through the Douyin web version.
As an entertainment app born from a culture of “short, flat, fast” and “3-second attention span,” Douyin’s move seems to signal a break away from “traffic first” logic and a pursuit of more long-term value accumulation. This change can also be seen with the launch of “Douyin’s Selected Content” in September 2024.
While introducing more original and high-quality content, Douyin actively leverages AI search technology to integrate information from across the web and focuses on consulting content. Users can click on “AI – Selected Consultations” in the “Trending Topics” section at the top of the Douyin app to view real-time featured news summaries and content summaries captured and generated by AI. Behind this design, Douyin clearly wants to use AI to carry and deliver longer-form content more efficiently.
Over the past two years, ByteDance has introduced AI technology to almost all content formats. Qishui Music introduced AI into the music creation process, Hongguo Short Dramas used AI to generate comic drama content, and Fanqie Novel incorporated AI into online writing. From news and music to novels and drama, ByteDance’s continued investment and promotion systematically leverages AI to create and distribute content.
01 From “short” to “long”
In September 2025, Douyin conducted a small internal test of its long-form writing feature, supporting high-quality text and image creation activities with a minimum of 300 words and a maximum of 4000 words. After the official launch, the word limit was further expanded to 8000 words. On the home page, these long articles appear in the form of images and scrolling text, accompanied by gentle music. The user is below the text[記事全文を読む]Click to access the complete content page.
Many media organizations and individual creators are now starting to experiment with this feature, and the content being published is primarily focused on in-depth stories, social news, and personal growth, which are relatively long-term cycles and not direct narrative directions.
Douyin wasn’t the only platform looking to expand its content boundaries last year. In July 2025, Xiaohongshu launched an innovative “convert long text to image” mode. The platform offers one-click AI typesetting, generates summaries and covers, and finally automatically splits 1,000 words of content into images that can be swiped to the right for publication, expertly adapting detailed content to users’ unique swiping habits for text and image notes.
Douyin and Xiaohongshu’s continued emphasis on long text and images essentially competes for high-quality creators and user dwell time. The former chooses to expand horizontally by inserting long-form text and image content of completely different styles into short video streams. The latter uses AI tools to dive deep vertically into existing text and image forms, lowering the barrier to content creation. At a deeper level, this is also an inevitable choice for content platforms after entering the era of stock competition.
In fact, as an ancient content medium, long-form writing has always had value.
The Attention Game, a 2024 research report from general marketing agency Mavericks, dispels common misconceptions. Although short-form content is widely popular due to its convenience, it is far less effective at building trust than long-form content. The report also noted that 34% of respondents believe long-form writing is more trustworthy and engaging than any other format. Despite being easier to view, short videos or short-form text content ranks lowest across all metrics. Only 10-12% of respondents approve of it in terms of factual accuracy, quality of presentation, and relevance of content.
On the other hand, it is also clear that Douyin’s move is aimed at competing with the complete ecosystem built by WeChat. As of October 2025, Douyin’s monthly active user base is 907 million, while WeChat’s monthly active user base is over 1.4 billion, according to QuestMobile data. There are still gaps in the number of users. Currently, Douyin’s monthly active users rank second among all networks behind WeChat.
As the traffic growth ceiling slowly approaches, Douyin may have realized that short videos are no longer enough to support the next stage of expansion. We need to look for new forms of content outside of our existing systems to fill in the long-missing pieces of the puzzle.
In a social space dominated by WeChat, Douyin clearly hasn’t given up on its layout. In March 2024, Douyin launched the “Co – raise an Elf” feature. As long as the two friends continue to message each other for three days, the “Spark” logo will light up. After sending messages for 6 days in a row, they can invite each other to raise a “little firefighter” together. As the number of interaction days increases, the spark value and Little Fireman’s level will increase at the same time. If there is no interaction for more than 24 hours, the spark and small firefighter will turn gray and may even disappear completely if interaction is not restored within the specified time.
This design quickly became popular among young users. To this day, raising little firefighters has become a trendy, ceremonial social activity, giving rise to Internet catchphrases like “You can argue, but you can’t extinguish the spark.” By quantifying social relationships and linking them to loss costs, Douyin significantly increased the frequency of interaction and emotional investment between users. This mechanism of increasing social stickiness has actually attracted many young users and, to some extent, inspired them to shift the focus of their daily chats from WeChat to Douyin.
After successfully “maintaining” social relationships, Douyin’s next step is to try to accommodate deeper content expressions. “People who read news don’t like to watch videos. They prefer to read articles,” explains a ByteDance insider. In a sense, this move is equivalent to embedding Toutiao within Douyin, offering more content formats to attract potential users.
Another reality is that Toutiao’s growth in size has stalled. After a long-term repurposing with short videos, the upper limits of pure news-based products became increasingly apparent. Rather than continue to invest in a product with no visible cap, it is better to place long-form content on Douyin and reallocate user attention with a larger traffic pool.
Just as WeChat has gradually built its official accounts and video channels based on social relationships, Douyin is also using its powerful algorithmic capabilities to expand the boundaries of its products. However, unlike WeChat, which naturally supports scenario reading, in Douyin’s highly entertainment-oriented ecosystem, it remains to be seen whether these knowledge-oriented and depth-oriented long-form contents can really capture the hearts of users.
02 News and information behind AI
In addition to relying on Douyin traffic, ByteDance continues to use AI technology to enhance its traditional business.
Take Qishui Music as an example. In addition to using Douyin for traffic and exploring innovative merchandising models, the platform has also actively built an “AI + music” creation ecosystem. Qishui Music launched the “Qishui AI Music Creation Laboratory” in November 2025. It helps creators simplify the process of writing, composing, recording, and arranging through AI tools, effectively lowering the threshold of music creation and enabling batch content production. By January 2026, Qishui Music’s monthly active users reached nearly 140 million, ranking 4th in the industry. In terms of daily active users, it ranked third after Kugou Music and QQ Music, and even surpassed NetEase Cloud Music.
In the long-form content space, ByteDance continues its “AI+” approach. Now, Douyin’s Hot News section allows AI to capture the latest news hotspots around the web 24 hours a day, integrate video and text information, and generate intelligent summaries.
These news items usually come from mainstream and authoritative media. Based on efficient capture, AI will further classify and aggregate. You can also ask questions about the information or ask the AI for interpretation by clicking the “Continue Questions” button at the bottom of the article.
ByteDance insiders believe that compared to traditional news articles, the advantage of AI-based consultations is that “users can read the article or click to view related videos, making the content consumption path more complete.” He also revealed to 36Kr that this kind of news and information “created” by AI will soon be populated into Douyin’s news feed. In other words, these AI-generated news items compete with the original long-form content in the same traffic pool.
But beyond the reading experience, more fundamental but under-discussed questions emerge. Can news and information be easily captured, disassembled, and redistributed? Where should the boundaries be as AI begins to participate in content creation, especially news content creation?
In fact, in current practice, AI cannot obtain first-hand information on the ground, so the way it handles news is nothing more than “secondary integration” rather than actual content production. However, this type of integration relies on algorithmic relevance rather than real-world fact checking.
When this technology is applied to the news field, responsibility becomes blurred. Previously, information compilation and fact-checking were the professional responsibilities of news organizations. However, in the AI era, even if there is a mistake in judgment, the AI is not held responsible, and the platform is often at fault for technical issues or information references. However, this process could lead to an even more rapid erosion of public trust in the news.
In other words, this content structure, which has high information density but low verifiability, itself creates new cognitive risks. In February 2025, 14 mainstream media organizations, including Forbes, Condé Nast, and the Los Angeles Times, filed suit against a company named Kohia, accusing it of mass copying articles from websites without permission to train models and create news summaries, and that the process tends to multiply and amplify the risk of “illusions” that damage media and publisher reputations.
On the other hand, the value of information itself continues to decline. In May 2025, the Graphite research team discovered that more than half of the content on the internet is generated by AI. AI can reorganize and distribute existing information infinitely, so information density increases rapidly, but improving the quality of information is difficult. On the contrary, the communication environment becomes noisy. People’s attention is further consumed and more judgment is required to avoid being flooded with misinformation and junk information.
In the age of information explosion, we need to guide and inform people about which information deserves more attention and which noise can be ignored. Just like in times of great material wealth, people can shop through any channel, but they still prefer to go to Costco and spend. Costco doesn’t have much choice, but every item on the shelf has been carefully selected.
