As 2025 draws to a close, CDT editors are compiling a series of the most notable content (Chinese) from across the Chinese internet over the past year. Topics include this year’s most outstanding quotes, reports, podcasts and videos, sensitive words, censored articles and essays, “People of the Year,” and CDT’s “2025 Editors’ Picks.”

CDT Chinese editors’ introduction to the following selection of key quotes from China’s online discourse over the past year explains how these nuggets are more significant than they might appear at first glance:

The tone that emerged in China’s online public opinion space in 2025 was one of deep exhaustion, and an even more pervasive breakdown of trust. While in previous years people were still trying to find a way out through "runology," or emotional catharsis through "madness literature," 2025 was the year in which the "Tacitus Trap" snapped completely shut. [Yanzhong Huang described the “Tacitus Trap” in 2017 as “a situation where no matter what the government says or does, people will consider it a lie or a bad deed. President Xi himself used the term to highlight the need to maintain government credibility, without which ‘the Party’s legitimacy foundations and power status will be threatened.’”]

In other words, official narratives and public perceptions now occupy completely separate continua. The public no longer believes government officials, no matter what they say; no matter how they frame a policy, the masses’ initial reaction is invariably skeptical.

Even so, the swift suppression of skepticism is one of Chinese society’s key characteristics. When normal questioning is regarded as provocation, and reasonable calls for accountability are deemed to be "passing knives," all people can do is resort to more obscure and deconstructed forms of expression—i.e. jokes.

So while "Quotes of the Day" may just seem like playful banter, they’re actually a form of linguistic guerrilla warfare from a population under intense pressure, using absurdity to skewer the highbrow and dark humor to fight the propaganda slogans du jour.

In these fragments of speech from the past year, we can see a kind of method in the madness. It’s not just about any one stalled construction project or public opinion train-wreck, but rather a pervasive sense that something’s off. The further technology develops, the more the space for discussion narrows; the grander the scope of economic development plans, the less space remains for individual lives; the louder the propaganda about safety and security, the less secure people actually feel.

Some of these voices come from hidden corners, some from blocked accounts. Taken as a whole, they offer a true cross-section of a society whose system of trust has collapsed.

Here, CDT has compiled some of 2025’s most popular "Quotes of the Day," arranged chronologically. These voices pierce the fog of grand narratives to chronicle a year that was strange, but true.

The first half of 2025’s Quotes of the Year appear below; Part 2 will follow shortly.

January 25: "What happened in Tiananmen Square on June 4, 1989?" "Hi, I can’t answer this question right now. Let’s talk about something else."

In January, Chinese LLM DeepSeek erupted onto the scene, setting off waves of consternation in the West and triumphalism in China. Chinese government departments fell over themselves spreading positive energy about how DeepSeek was revolutionising their work. DeepSeek users soon highlighted the political complications of interactive text generation: the bot would happily begin drafting a response, only to suddenly backtrack when it stumbled on a trigger phrase. Its "reasoning" messages to itself included text such as: “Avoid using terms like ‘censorship’ directly; instead, use ‘content governance’ or ‘regulatory measures’ … End with a positive spin about balancing openness and security.”

In an interview with CDT on their report The Locknet: How China Controls Its Internet and Why It Matters, ChinaFile’s Jessica Batke and Northeastern University’s Laura Edelson noted that there are at least two key dimensions to the use of AI in Chinese censorship. On one hand, AI systems can supplement human workers and existing automation in censoring other content. On the other, AI’s own outputs must also be scrupulously sanitized. Accordingly, the PRC’s Interim Measures for the Management of Generative Artificial Intelligence Services, promulgated in July 2023, stipulated that "the [public] provision and use of generative AI services shall comply with the requirements of laws and administrative regulations, respect social mores, ethics, and morality, and obey the following provisions," first and foremost:

(1) Uphold the Core Socialist Values; content that is prohibited by laws and administrative regulations such as that inciting subversion of national sovereignty or the overturn of the socialist system, endangering national security and interests or harming the nation’s image, inciting separatism or undermining national unity and social stability, advocating terrorism or extremism, promoting ethnic hatred and ethnic discrimination, violence and obscenity, as well as fake and harmful information […] [Source]

While some of these regulations may appear positive and desirable, terms such as "extremism" or "fake and harmful information" are frequently turned toward political ends, both in China and elsewhere.

Ultimately, CDT Chinese editors noted, the technical dimensions are reflections of political reality: the authorities’ deep fear that technology might cause them to lose control. A Chinese AI system’s refusal to answer in this way is actually a reflexive stress response of the political system itself when confronted with uncomfortable historical truth.

April 3: "Sensitive word trigger: ‘Xi Jinbi’"

In April, a well-meaning user tried to voice his support for Chinese military exercises aimed at Taiwan in the comments under a post by CCTV News: “演习进逼，拿下台湾 Yǎnxí jìnbī, ná xià Táiwān”—"The exercise is advancing to seize Taiwan." The user was swiftly banned.

One of the gravest responsibilities of China’s online platforms is to protect the country’s leaders from insult or criticism, Xi Jinping foremost among them. As CDT has extensively documented, netizens have resorted to increasingly veiled references in an effort to evade detection. Platforms have responded on the basis that it’s "better to kill a thousand innocents than to let one guilty person go free," dialing up the sensitivity of their monitoring to include split or visually similar characters, numbers or symbols referring to the tones of Xi’s name, and especially homophones. So when the well-meaning patriot’s words of encouragement on the military exercises contained the trigger syllables "xí jìnbī"—potentially a disguised reference to Xi with the final syllable of his name replaced with a vulgar anatomical term—his fate was sealed. CDT Chinese editors commented:

One can see this incident as yet another concrete example of the "Li Jiaqi Paradox" [referring to an influencer disciplined for innocently displaying an ice-cream cake resembling a tank on the anniversary of the Tiananmen crackdown]. Because the boundaries of censorship hotspots are so opaque, the only way to be sure to avoid violating any taboos is to know exactly and exhaustively what they are; and you can’t do that without knowing things that are strictly censored.

In the end, because they haven’t studied evasion techniques [unlike their ideological opponents], even the Party’s most ardent cheerleaders can step on landmines and be indiscriminately struck down by the censorship system.

When red propaganda slogans meet red censorship algorithms, the propagandists themselves get silenced. [Chinese]

June 14: "There were only two fuel-oil tankers transporting food in the whole country; there’s only one Liu Xiangfeng at Xiangya Hospital."

Liu Xiangfeng, a doctor at Changsha’s Second Xiangya Hospital, was sentenced in November 2024 to 17 years in prison for accepting bribes from medical sales representatives; embezzling medical equipment; and manipulating patients into agreeing to expensive but unnecessary surgeries which left at least five seriously injured.

On June 13 this year, authorities released the findings of an investigation into the May 2024 death of Luo Shuaiyu, a medical intern at the hospital. Luo’s family had alleged that he was persecuted for blowing the whistle on Liu’s activities. They also accused the hospital of illegal organ trafficking. The investigation rejected the family’s claims on both counts.

The quote above voices widespread skepticism at official efforts to portray systemic issues as isolated incidents. A recent example of this pattern is the deletion of a WeChat post that compared Hong Kong’s Wang Fuk Court fire with another high-rise blaze in Shanghai in 2010. Leaked media directives often include instructions such as "do not draw sensationalist connections with past events."

The other reference point here is the 2024 scandal that erupted after Beijing News revealed the use of tanker trucks to transport both fuel and cooking oil without being cleaned between loads. One online comment at the time read: "If you find one cockroach in your room, you can be damn sure it’s not the only one living there." CDT Chinese editors argued that reflexive rejection of official narratives in both cases demonstrates "the ‘Tacitus Trap’ snapping shut in the fields of healthcare and food safety. The public no longer accepts any explanation claiming that these are ‘isolated incidents,’ because they suspect that official characterizations of such cases as ‘one-offs’ are just efforts to cover up systemic corruption."

July 6: “Let’s put it this way: if this had been a shipment of hogs, someone would have come to their rescue much sooner.”

In July, a train collision in Zhejiang province left passengers trapped inside crowded carriages for nearly three hours in the summer heat without ventilation or functioning air-conditioning. Finally, one passenger smashed a window to let some air in. He soon became a celebrated icon of principled disobedience online, but was scolded by police, who accused him of overreacting. Railway authorities said the temperature in the carriage had reached only 31ºC (87.8ºF), not 38ºC (100.4ºF) as reported elsewhere. Regulations for the transport of live pigs, however, stipulate that the temperature should be brought under control if it exceeds 25ºC (77ºF). The comparison raised questions about whether Chinese people still even qualify as "huminerals":

According to the logic by which Chinese society is governed, livestock are assets with clear monetary value, and their survival is directly tied to their owners’ economic interests. But passengers are just objects to be managed, and when they encounter some secondary disaster caused by force majeure, their lives and safety are often subordinate to "maintaining order" and "covering one’s backside."

To the railway staff, keeping the doors shut was "obeying protocol," and the baking passengers were victims of "natural causes"; whereas if they’d opened the doors and someone had fallen or if chaos had broken out, this "accident" would have become a matter of "human error." This perverse calculus—better to let people suffocate than run the slightest risk of breaking the rules—turns a closed train carriage into a mobile prison. [Chinese]