The Cyberspace Administration of China said on 2 September that its campaign against the misuse of AI had removed more than 5.61 million items of illegal information [1].
It said more than 49,000 accounts had been dealt with, and more than 2,400 websites and apps [1].
The second stage focuses on AI-generated false information, violent and vulgar content, impersonating other people, and content that harms minors, according to the regulator [1].
What it covers
Content and services on Chinese platforms. The release names Douyin, Kuaishou, Weibo, Tencent, Baidu, Bilibili, Xiaohongshu, Zhihu, Douban and Taobao as having tightened their moderation [1].
It says the chatbots Doubao, Yuanbao, Qianwen and Ernie Bot restricted rule-breaking output and applied its labelling rules for AI-generated content [1].
Its examples include accounts using face-swapping and voice cloning to impersonate public figures, and AI-generated videos showing children in harmful situations [1].
The four-month campaign was announced on 30 April, in two stages: the first on AI services themselves, such as unregistered models and missing content labels [2].
What it does not tell you
How much of the removed content was made with AI, or how many items were judged illegal rather than merely against platform rules. The release gives totals only [1].
What the regulator counts as illegal. Decisions are made under Chinese law and platform agreements, with no independent review described [1].
Comments