Welcome to the HRIC Digital Rights Report, our monthly roundup of developments in China’s digital rights landscape, including cybersecurity, AI, digital transnational repression, censorship, the Great Firewall, and internet freedom.
This Month: Flooding Tragedy, Social Stability & Censorship
Devastating flash floods along the Nepal–Tibet border in late August exposed the human cost of China’s information controls. While more than 1,000 people were confirmed dead and thousands remained missing across the affected region, Chinese authorities minimized information about casualties, restricted independent reporting, and detained people for sharing information about the disaster.
Eyewitness footage disappeared from Chinese social media, while authorities used image-recognition technology to identify and remove sensitive content and flooded online spaces with official material to bury independent accounts. Authorities reportedly detained or fined at least 10 people for questioning official accounts of the disaster online.
This response reflected a broader pattern in China of prioritizing “social stability” over the free flow of information. Under this approach, authorities restrict information that could generate public criticism, expose official failures, or lead to demands for accountability. At the same time, state media promote carefully managed narratives that emphasize social order and the government’s ability to respond to crises. Following the Nepal–Tibet floods, for example, Chinese state media focused on rescue operations and the government’s response while downplaying the full scale of the destruction. Similarly, five years after the deadly 2021 Zhengzhou floods, in August, authorities continued censoring posts that commemorated victims and preserved public memories of the disaster.
This approach also extended beyond disasters and explicitly political speech. Authorities removed a straightforward WeChat post describing the exhausting security checks required to visit Beijing’s Zhengyangmen gate, illustrating how even ordinary frustration can be censored when it reflects negatively on the government’s security agenda. Similar restrictions affected expressions of grief in Hong Kong, where organizers encountered rejected advertisements and shuttered memorial displays after hundreds gathered to mourn Wang Wang, a stray dog tortured and killed in Guangdong.
Although such censorship may protect the government from criticism and preserve an appearance of stability in the short term, it comes at the expense of public safety and access to reliable information. During a disaster, timely reporting can help residents understand the risks they face, locate missing relatives, identify areas in need of assistance, and make informed decisions about whether to evacuate. Restricting that information can delay emergency responses, fuel rumors, weaken public trust, and make it more difficult to determine the true scale of the damage.
The effects of these controls are not limited to emergencies. A WeChat essay translated by China Digital Times this month described China’s keyword-based censorship as blunt and counterproductive, arguing that it blocks harmless speech, fails to address genuine harms, and forces users to communicate through riddles. The essay highlights how broad and automated restrictions can interfere with ordinary communication, even when users are not criticizing the government or discussing politically sensitive issues.
The examples covered this month—from disaster footage and complaints about security checks to memorials for a stray dog—perfectly illustrate this problem. They show that speech does not need to resemble conventional political dissent to become a target in China. Censorship increasingly shapes not only what people may criticize, but also what they may know, witness, mourn, and say about their everyday lives. As the boundaries of restricted speech continue to expand, the distinction between controlling political dissent and controlling ordinary public expression becomes increasingly difficult to draw.
In Case You Missed It: Other Digital Rights and Cybersecurity News
Great Firewall, Censorship, and Information Control
Translation: “You Can’t Address Sexual Assault at Banquets Without Addressing the Nature of Power”: A censored WeChat post examined two recent cases of alleged sexual assault by Chinese officials, arguing that piecemeal responses to individual scandals fail to address the systemic issue: the lack of institutional checks on official power that enables such abuse to recur.
Translations: “Sometimes Forgetting Is as Terrible as the Disaster.” Censored Reflections on the Deadly Zhengzhou Floods, Five Years On: Marking five years since Zhengzhou’s deadly 2021 floods (398 dead, many deaths concealed), two censored Chinese essays reflect on memory, grief, and the “Raincoat Dad” who mourned his drowned daughter, warning that societal amnesia allows man-made disasters to recur. Posts merely commemorating the tragedy, without demanding accountability, were still censored.
Translation: No Scaling the Great Firewall in the Library: A Hunan Library notice suspended library Wi-Fi after patrons were caught using censorship circumvention tools, publishing partly redacted phone numbers of offenders as a warning that internet use is tied to real identities. Commenters noted circumventing on personal mobile data is safer than on institutional networks, which face heavier scrutiny and retaliation against administrators.
HRIC on Twitter/X: There has been a recent crackdown on content referencing a traditional bamboo “cicada” toy as satirical commentary on Huawei. The toy, which emits a continuous “wah-wah” sound when shaken, was noticed by netizens for its sound effect similar to the synchronized, exaggerated gasps of surprise from the audience at Huawei launch events. Searches for “bamboo cicada” on Weibo were wiped clean, while Douyin, Kuaishou, Xiaohongshu, and WeChat public accounts saw a flood of account restriction notices. On August 2, “Is it overkill to take down Yu Chengdong parody videos?” trended on hot searches, with some state-affiliated media starting to defend the removals and bans.
AI Propaganda and Social Media Manipulation
Hanging a Banner for Chinese AI: People’s Daily recently ran a symmetrical, parallel-column front page proclaiming Xi Jinping’s “concern” for both domestic AI development (”beneficial, safe, equitable”) and China’s role in bridging the global AI divide for developing nations. CMP explains: “The unusual column design, with its carefully conceived symmetry—a propaganda appropriation of the traditional parallel saying, or duilian (对联)—is meant no doubt to be a powerful parallelism announcing the Chinese Communist Party leadership’s twin convictions on artificial intelligence, for the country and for the world.”
HRIC on Twitter/X: The Wall Street Journal reported that multiple new studies have found U.S. artificial intelligence models unconsciously exhibit similarities to China’s censored chatbots when answering certain sensitive questions, while AI companies have made limited efforts to address this issue.
The Problem with Peach: Chinese AI-generated influencer “Peach Fang,” star of a viral AI drama with 200 million views, sparked controversy after promoting contact lenses she cannot physically wear or test, raising questions about AI personas making false claims to bodily experience. State media and lawyers noted AI “endorsers” lack legal personhood, exposing regulatory blind spots around synthetic marketing.
Translation: “AI Has Determined That My Press Card Is a Fake”: A China Youth Daily journalist recounts the surreal experience of an interview subject’s AI chatbot confidently declaring his official press card a fake, illustrating the growing, sometimes absurd, ways AI tools are reshaping trust, misinformation, and censorship dynamics in Chinese journalism and society.
Cybersecurity & Data Privacy
HRIC on Twitter/X: On August 5, U.S. cybersecurity company VulnCheck released a research report stating that routers shipped from Shenzhen Zhibotong Electronics (Zbtlink/ZBT) come pre-installed with a backdoor program in their firmware. As soon as the device is powered on, it continuously initiates outbound connections to several fixed overseas command-and-control servers, with two primary nodes hosted on Alibaba Cloud in Shanghai and Shenzhen, respectively.
Royal Navy drone cameras linked to China raise security concerns: Cameras on the UK Royal Navy’s Chinese-component K3 Scout surveillance drones, used by UK special forces, were found sending routine “heartbeat” signals to a device in China during a cybersecurity assessment. The UK Ministry of Defence stated that no sensitive data was compromised but has cut the drones’ internet connectivity. The incident adds to mounting concern over Chinese-made components embedded in Western military and critical infrastructure systems.
Hong Kong mulls AI facial recognition-enabled drones to police smoking ban at construction sites, 39 fines issued so far: Hong Kong’s Labour Department has issued 39 fines and 11 improvement notices since a construction-site smoking ban took effect in mid-July, and is now exploring adding AI facial recognition to its infrared drones (already used to detect cigarette heat signatures) to better identify offenders, with development expected to take 9–12 months. The move raises questions regarding the expanded use of AI facial recognition drones for surveillance purposes.
Hong Kong police arrest 10 for dangerous driving in first drone-assisted traffic crackdown: Hong Kong police arrested 10 drivers for dangerous driving on the Kwun Tong Bypass in their first drone-assisted traffic operation, using aerial surveillance alongside unmarked “stealth cars” to gather evidence.
HRIC on Twitter/X: A New York Times’ investigation into a recent leak from cybersecurity researcher Mark Hofner on the “Foreign Personnel Dynamic Control Platform” reveals new details about the tool, which tracked over 700 foreign nationals residing locally. Greg Walton, a senior investigator at the Canadian SecDev Group, described the incident as a symptom of “surveillance proliferation” driven by an ever-expanding ecosystem of vendors, contractors, and public security departments: each additional platform creates more places where sensitive personal data could be misconfigured, copied, or exposed.
Digital Transnational Repression
Toolkit to Counter the Chinese Communist Party’s Transnational Repression: Human Rights Foundation’s new anti-TNR toolkit includes tips for digital safety for those impacted or potentially targeted by transnational digital repression.
Taiwan says it was hit by ‘abnormal’ AI-assisted cyber-attack: Taiwan detected a first-of-its-kind cyberattack on government agencies last month, in which hackers used open-source AI agents like OpenClaw to build an autonomous hacking tool that operated like a coordinated cyber team. Taiwan didn’t name a culprit, but has long accused China of “hybrid warfare” including near-daily cyberattacks.
A fake website and a deluge of CVs: the Australian firm embroiled in an FBI probe into alleged Chinese espionage: A small Brisbane consultancy, Horizzen, discovered a fake AI-generated website impersonating its identity had been used as part of a broader Chinese intelligence recruitment network targeting Western defense and security experts.
Civil Society & Resistance Under Pressure
From petitions to memes: How students made dissent visible: The piece traces how Chinese student dissent has evolved from formal petitions toward subtler, coded forms like memes, puns, and symbols (echoing tactics from the 2022 “White Paper Movement”) to evade censorship while still expressing discontent. This shift reflects both technology’s role in activism and authorities’ tightening crackdown on any direct challenge to the state.
Technological Competition & State Power
Nvidia senior manager linked to Supermicro scheme smuggling AI servers to China: Taiwan indicted nine people, including a senior Nvidia manager and two Supermicro employees, for allegedly forging documents to smuggle 74 AI servers with restricted B300 chips to China via Japan, Indonesia, and Hong Kong, evading US export controls.
The rise of AI weather forecasting as China emerges as a leader in race to improve predictions: As Typhoon Dolphin approached, Chinese AI weather models like Shanghai AI Lab’s Fengwu, Huawei’s Pangu, and Fudan’s Fuxi worked alongside traditional forecasting, with Fengwu predicting the typhoon’s landfall time and location within 30 minutes and 30km five days out. China is emerging as a leader in AI weather prediction, though such models still lag traditional systems in forecasting storm intensity.
ByteDance trains massive AI model in bid to rival Anthropic: ByteDance is pre-training a roughly 10-trillion-parameter AI model — more than triple the size of Moonshot’s Kimi K3 — aiming to approach the scale of Anthropic’s Mythos system, per industry estimates. The project reflects intensifying competition among Chinese AI labs to match or exceed US frontier models, amid earlier US accusations that Moonshot’s Kimi K3 was built by distilling Anthropic’s technology.
HRIC on Twitter/X: Forbes reported on August 5 that U.S. AI data labeling companies are supplying OpenAI, Anthropic, and even the U.S. military on one hand, while selling the same sets of training data to China’s top AI labs on the other. For years, Washington has imposed layer upon layer of restrictions on chips, yet has done almost nothing to regulate training data.
Taiwan detains Nvidia worker in chip smuggling probe: source familiar with case: Taiwanese prosecutors detained an Nvidia employee as part of a widening probe into smuggling roughly 50 AI servers with restricted Nvidia chips to China via forged documents, bringing the total detained to seven people including Super Micro and Albatron staff. Nvidia denies involvement, saying it only sells through vetted partners and “smuggling is a nonstarter.”
China’s tech advances are causing chaos from Silicon Valley to the White House: A wave of Chinese breakthroughs—free, powerful open-weight AI models like Kimi K3, mass-produced specialty chips, and advanced robotics—has triggered market selloffs and split Silicon Valley, with firms like Microsoft and Nvidia opposing restrictions on open models while OpenAI and Anthropic call them security risks.
Current Events
Could AI take your job? Some workers in China already know the answer: Chinese workers, from Wuhan taxi drivers displaced by autonomous vehicles to a Beijing cinematographer replaced by AI software, describe deep anxiety over AI-driven job losses, with some winning court-ordered severance after AI replaced them. Even China’s state media now acknowledges the “collision” between AI and traditional labor rights, signaling a shift in Beijing’s tone on the issue.