Why Is Google Translate Suddenly Failing in Mainland China?

In the past few days, users in mainland China discovered that Google Translate no longer works.

The service that should be provided by translate.google.cn has been replaced with a screenshot of a "search page".

Any click on the page redirects to Google Translate’s Hong Kong domain translate.google.hk, which is inaccessible for mainland users.

As a result, Chrome’s built‑in translation feature and other software that rely on the same API, such as the document viewer KOReader, have stopped working.

Google officially withdrew from the Chinese mainland market in 2010.

Former Minister of Industry and Information Technology Li Yizhong once said that foreign internet services must comply with Chinese law to operate in the country.

Reports indicated that Google attempted to relaunch its search service in China in 2018 and 2019 as part of the "Dragonfly" project, which would have censored search results and logged users’ locations and browsing history. The project was later abandoned after internal conflicts within Google’s privacy team.

In November of the previous year, some users reported that Google Translate rendered certain AIDS‑related terms with offensive Chinese translations, sparking outrage. Google later explained that the issue was caused by a "mode" and was fixed.

Analysts argue that much of the Chinese corpus on the international internet is not controlled by Chinese speakers, and that biased or "dirty" training data can lead to inaccurate translations. They also point out gaps in model‑to‑API management.

Why do certain obscure Chinese terms get translated into hostile language toward China when the source language is forced to English? The answer lies in flawed training data and inadequate oversight.

Reference: https://tcrn.ch/3E6HC3b