IE like Crypto AG:
In 2020, it was revealed that the Swiss company, Crypto AG, which provided secure communications services to ~120 governments throughout the 20th century, was secretly ran by the CIA and West German Intelligence. The CIA and later NSA were able to read encrypted communications for many countries such as Saudi Arabia, Iran, Italy, Indonesia, Iraq, Libya, Jordan and South Korea.
Look what autocratic(ising) governments don’t do shit against. And what their opposing governments tell about the autocrat(ising) ones.
Those are more likely to be honeypots imho, as they claim to be one thing but likely work for another.
COINTELPRO (or as I’d rather label it, SOWTERROR), showed that the US government heavily opposed mutual aid and class solidarity, and combatted the Black Panthers, by defaming them, inserting gun control laws once they started police watching, bombing their homes and so on.
That said… let’s take an example. Correct me if I’m wrong, I’m working from my memory here.
DeepSeek is open-source, while ChatGPT isn’t. DeepSeek does however, participate in censorship (which Muskrat’s models also do, just in a different direction). DeepSeek is much more efficient, threatening Western models.
And imho, the latter is why many Western governments oppose using it. They say it’s spyware, but how can we be sure ChatGPT isn’t? It’s proprietary shit.
But both models are honeypots, just in their own ways. Sure, DeepSeek is open-source as it claims. But it also censors itself to unnecessary extents. Then there’s ChatGPT, which claims to be secure… but how can we know that when it’s proprietary and we cannot crosscheck its source code?
DeepSeek the service censors, but you can run it yourself uncensored.