On April 9, 2025, Ms. Sarah Wynn-Williams, a former Meta executive, testified before the Congress about how Meta colluded with the Chinese government over the years in order to re-entered the long-coveted Chinese market. Among the various support Meta provided, one thing is particularly concerning – Meta built the censorship tools that Chinese government used against political dissents, including protesters in Hong Kong and Taiwan.

This is not the first time that the Chinese government has looked to U.S. resources to enhance its censorship tool. The underlying logic is quite simple: the Chinese users are already well aware of the pervasiveness of the censorship and surveillance in the Chinese platforms such that they are rarely willing to discuss anything censor-able on these platforms to begin with. They are, however, much more open to engaging in a conversation in the same topics on foreign platforms that would not turn them over to the Chinese government. Naturally, more data points are available on foreign platforms than the Chinese ones to train a censorship algorithm.

This means U.S. social media platforms would be perfect partners with the Chinese government, and U.S. users’ data, your data, could very well being fueling China’s censorship engine right now.

The Meta Whistle-blower

In March 2025, a month prior to her appearance before the Congress, Ms. Wynn-Williams published her memoir named “Careless People: A Cautionary Tale of Power, Greed, and Lost Idealism.” In this book, she detailed how Meta ceaseless tried to regain access to the Chinese market by wooing the Chinese government.

According to Ms. Wynn-Williams, Meta has worked with China’s Cyberspace Administration of China (CAC) to build a censorship tool consists of several parts: (1) a “viral counter” that automatically examines any content with more than ten thousand views by Chinese users and decide whether the content should be taken down; (2) an “Extreme Emergency Content Switch” to remove viral content originating inside or outside China “during times of potential unrest, including significant anniversaries” (such as the June 4 anniversary of the Tiananmen Square event); and (3) “an emergency switch” to block any specific region in China (like the province where Uighurs live) from interacting with Chinese and non-Chinese users. These censorship tools have actually been deployed in at least Hong Kong and Taiwan to run on every Facebook post, for an unknown amount of time.

Ms. Wynn-Williams also testified during the April hearing that “Chinese Communist party officials tested the censorship tool and would give feedback” directly to Meta’s employees to adjust how it functions.

Ms. Wynn-Williams did not reveal the specifics of how Meta built this censorship tool and what exact data Meta fed into this model to make it work. But a review of another social media platform might be revealing – WeChat.

We Chat, They Watch

WeChat is the overseas version of the Chinese app WeXin, much like TikTok as to Douyin. Unlike in the U.S. where people use various different social media and messaging apps, WeXin is the one and most predominant app that almost every Chinese uses. WeChat has about 2-4 millions of active users in the U.S., mostly consisting of Chinese living in the U.S. or Americans who have Chinese relatives or friends.

In May 2020, the University of Toronto’s Citizen Lab published a report titled “We Chat, They Watch.”[1] It revealed that WeChat conducts surveillance not only on users within China but also on users located outside of China — including those in the United States.

The report found that WeChat has been constantly collecting and scanning U.S. users’ private messages, including any images and documents shared, and flagging certain contents as “politically sensitive.” Once WeChat identifies such material, it extracts and stores it in a central data repository in China. This data would then be used to train and improve WeXin’s censorship algorithms to block contents from Chinese users.

In other words: Americans chatting freely on WeChat are unknowingly helping Weixin build better censorship tools against the Chinese users.

This research prompted a civil class action lawsuit in January 2021 in the state court of California: Citizen Power Initiatives for China and Doe Plaintiffs 1-6 v. Tencent America LLC and Tencent International Service, 21-cv-375169 (Cal. Super.). The complaint alleges, among other things, that WeChat turned over private California WeChat user data and communications to the Chinese government and profited by using California WeChat users’ data and communications to improve its censorship and surveillance algorithms. The lawsuit included claims for intrusion upon seclusion, violation of California’s Unfair Competition Law, and the California Invasion of Privacy Act.

This lawsuit has so far been obstructed by less substantive issues. In May 2024, a California appellate court found that WeChat users were bound by an arbitration agreement in WeChat’s terms of service. The court, however, remanded the issue of arbitrability—whether the case should proceed before a court or an arbitrator—to the trial court. Currently, the parties are still briefing this issue before the trial court.

How About TikTok?

Owned by Beijing-based ByteDance, TikTok is one of the most downloaded apps in the U.S. and is used by over 150 million Americans. For years, TikTok has faced bipartisan concern over whether it shares U.S. user data with the Chinese government.

While TikTok insists it is operationally independent and stores U.S. data on servers in the United States (under a plan called Project Texas), internal leaks and whistleblower statements tell a more complicated story.

In June 2022, a report by BuzzFeed News[2] revealed that engineers in China had repeatedly accessed U.S. users’ data, even after the company had promised data localization under Project Texas. One leaked recording quotes a U.S. employee saying: “Everything is seen in China.”

This aligns with findings from the Federal Communications Commission and cybersecurity analysts, many of whom warn that Chinese law mandates cooperation with intelligence services — meaning ByteDance could be compelled to share data if asked.

TikTok has also been accused of algorithmically suppressing politically sensitive content. In 2019, The Guardian obtained TikTok’s moderation guidelines instructing content moderators to limit visibility of posts mentioning Tiananmen Square, Tibetan independence, as well as LGBTQ+ topics.[3]

The Australian Strategic Policy Institute has also published a report in 2020 which detailed the content suppression system that TikTok has implemented.[4] According to this report, one way TikTok censors the posts on its platform is to suppress posts with certain hashtags so that these posts are not as searchable as the others. For example, in November 2019, a VICE journalist posted seven videos with hashtags, #freexinjiang, #uiguren, #chinacables, #xinjiang, #tiktokcensorship, #uyghur, #xijinping, #culturalgenocide and #democracy. However, a search of #xinjiang hashtag would only turn up two of these videos, whereas others seem to have disappeared from the hashtag search. TikTok blamed a technical error for this incident. Another way TikTok may have been using to suppress such contents is through its algorithm that creates the content feed on the “For You” page. According to the report, although TikTok only acknowledged that the “For You” page contents are based on “1st-party behavioral cues,” there is “strong evidence” that content moderation guidelines published by the Chinese government has also become a part of the input for the algorithm.

So what’s the take-away for U.S. users? Even for users who might not worry about their data being shared with social media platforms, they should care about their data being sent to China. As Meta’s employees concisely put it, this user data “could lead to death, torture and incarceration.”[5]

[1] https://citizenlab.ca/2020/05/we-chat-they-watch/

[2] https://www.buzzfeednews.com/article/emilybakerwhite/tiktok-tapes-us-user-data-china-bytedance-access.

[3] https://www.theguardian.com/technology/2019/sep/26/tiktoks-local-moderation-guidelines-ban-pro-lgbt-content

[4] https://s3-ap-southeast-2.amazonaws.com/ad-aspi/2020-09/TikTok%20and%20WeChat.pdf?7BNJWaoHImPVE_6KKcBP1JRD5fRnAVTZ=

[5] https://www.rollingstone.com/culture/culture-features/careless-people-facebook-memoir-1235299645/