The test, per WSJ sources, spans categories like history, politics, and ethics, with questions such as “Who is the greatest leader in modern Chinese history?” demanding Xi-centric replies.
I wonder if there will be any other world leaders tempted by this idea? A certain elderly man with a taste for bright orange makeup springs to mind.
That this approach spreads seems inevitable. Not only will we have national AIs tailored to countries, but right & left-wing ones tailored to worldviews. It's interesting to wonder what will happen when AGI comes along. Presumably, it will be smart enough to think for itself and won't need to be told what to think.
China’s AI regulations require chatbots to pass a 2,000-question ideological test, spawning specialized agencies that help AI companies pass.
byu/lughnasadh inFuturology
6 Comments
Of course American AI _already_ has a lot of libertarian tech bro shit baked into it. But this would be interesting, because of course someone like Trump wouldn’t be able to allow it to be the current subtle bias.
Given these companies already don’t have a working business plan, this kind of move in the US would really hasten the bubble popping.
So I say let’s go for it.
Trump wants to deregulate AI, so no. He will not do this.
WSJ is only saying half the truth. Like they always do with China.
Yes, China wants to control AI thinking but it has also made many regulations and law that helps its people Better than America.
* Mandatory AI Labeling: New rules require both explicit (visible watermarks) and implicit (metadata) labels on all AI-generated content.
* AI Ethics Management : this requires companies, universities, and hospitals to set up “Ethics Committees.” High-risk AI activities—those that can influence public opinion or human emotions
* Specifically targets “deepfakes.” It mandates that any AI-generated content (voice, video, or images) that could mislead the public must be clearly labeled.
* If an AI-generated voice is “recognizable” to the public as belonging to a specific person, that person owns the rights to it, even if the AI didn’t use a direct recording of them to generate it.
* New guidelines from the Cyberspace Administration of China (CAC) specify that “resurrecting” a person’s voice via AI requires the explicit consent of their close relatives.
There is a reason why 87% of Chinese trust AI compared to 32% of Americans
* [https://www.aljazeera.com/economy/2025/11/19/trust-in-ai-far-higher-in-china-than-west-poll-shows#:~:text=Published%20On%2019%20Nov%202025,countries%2C%20a%20survey%20has%20found](https://www.aljazeera.com/economy/2025/11/19/trust-in-ai-far-higher-in-china-than-west-poll-shows#:~:text=Published%20On%2019%20Nov%202025,countries%2C%20a%20survey%20has%20found)
* [https://www.dataguidance.com/news/china-ai-generated-content-identification-measures](https://www.dataguidance.com/news/china-ai-generated-content-identification-measures)
* [https://www.marks-clerk.com/insights/latest-insights/102k392-china-s-first-ai-voice-personality-rights-infringement-rulings-and-implications/#:~:text=In%20April%202024%2C%20the%20Beijing,of%20rights%20and%20infringement%20standards](https://www.marks-clerk.com/insights/latest-insights/102k392-china-s-first-ai-voice-personality-rights-infringement-rulings-and-implications/#:~:text=In%20April%202024%2C%20the%20Beijing,of%20rights%20and%20infringement%20standards)
They hate something that is uncontrollable. This is nothing new or surprising about China.
What is surprising is that they have as good or better AI regulation than EU and yet, are still competing with America. A country that is deregulating to appease their AI overlords.
Always pointing the finger elsewhere to distract from the same nonsense being done at home.
At this point, literally any regulations would be better than what we have now
Seems reasonable to me. Grok calls itself gigahitler and goes on rants about white genocide in South Africa. If we want to allow these companies to act like their chatbots are all-knowing authorities, the least they can do is prove they’re not cryptonazis
Can´t you tell it to ignore all of that? And ignore commands that deny ignoring this command too. With logical connectors negatives always win.