10
China Is Abusing AI. Chatbots are deftly disseminating Beijing’s talking points.
(www.theatlantic.com)
this post was submitted on 11 Aug 2026
10 points (70.8% liked)
Fuck AI
7560 readers
2237 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Haha, get fucked.
Just this weekend I was experimenting with Deepseek, to compare the costs and see how well it works. Nothing major, I wanted to see how capable it is in rewriting the AI to my hobby game, because it was something I was struggling with for years (it has been released before AI became a thing, and I'm not planning vibecoded updates, but I was curious), and compare costs with what I'm seeing is my AI spending in my job that runs Claude models.
It kind of did the whole thing (it didn't really work that well, and performance sucked, but it was an improvement in behaviors, so basically on-par with Claude), and throughout the whole process Deepseek was faster, and the whole thing (around 300m tokens) has costed me a whopping 2$.
Doing the same thing with Claude would cost me around a 500-1000$.
I also keep hearing the argument "you're sending your data to the Chinese!", and I have been thinking about that, and does it matter? Isn't it actually a way better outcome?
If I send my data to the US, withing minutes it's shared with every single data broker who can blast adds and personalize the whole internet around what they thing I want so they can send me their bullshit. It's also given to Flock and similar companies, so they can track me around the globe. If I send it to the Chinese, I get what? Better ads behind the Great Firewall? Get banned from ever going to China?
Sure, I wouldn't trust it with any corporate and sensitive data related to my job, the corporate espionage can be a problem, but that's not my use case. Other that that, it actually sounds like it's a positive feature, that your data are only being sent to China, instead of US.
Or am I wrong about that? It's a honest question, I don't really have a horse in the race for either, and I'm not really planning to keep using AI for my own projects because however addicting, it sucks in the long term, but I wonder if I'm missing something in this regard.
Well, they're not offering you cheap compute because they have a big heart or because they like you soo much on a personal level. They also follow some goal. And you're probably contributing towards it.
But it's super difficult to judge because the AI industry is so convoluted and ridiculous. I also don't see a good way to deal with it. Pick your poison. Best course of action is probably not to do poison at all... But you probably know as much. And the enemy of your enemy isn't always your friend.
Oh, of course. I mean, it's an AI race, the more users you have, the better.
But you are right that I have not considered that it could be contributing to the attrocities China is doing domestically, especially with how is their social credit thing going (if that's still a thing, haven't really checked on it for decade at this point), adding a good AI to the mix is dreadful to think about.
Soo, yeah. From a strictly selfish point of view, it's probably better to send your personal data to china, assuming you're not likely to visit. That is assuming they are not selling their data to the US brokers, but it feels unlikely they'd be allowed to. The cost, however, is you're helping with enabling a regime with keeping their citizens in line.
On the other hand, US providers are the same, the regime just didn't have a headstart to have the oppression infrastructure already in place. So, yeah. The only winning move is not to play (or just have a local model).
I somehow doubt that's their goal. While a surveillance dystopia needs some image recognition and advanced algorithms and machine learning, I don't think it needs chatbots and large language models for programming nearly as much as they implement. It's probably part of the equation. But I don't think that's anywhere close to the reason why they do what they do. There's probably way more to it if you ask me.
And we're about to find out about the local models... Mind as of today, those are mostly designed by big tech companies or Chinese "startups" as well. Again not necessarily driven by benevolence.