We have information that Moonshot distilled Fable for the development of K3

A fresh claim says Moonshot AI copied ideas from Anthropic’s newly released Fable system to help build its K3 model, allegedly using an internal setup designed to quietly pull from U.S. AI systems at scale and dodge detection. The report also says Moonshot bought powerful Nvidia server gear and tapped similar machines in Thailand to train its models. In plain English: the accusation is that a Chinese AI lab didn’t just “learn from” public work, but may have secretly siphoned off expensive American know-how.

But in the comments, the real fireworks were about whether anyone should believe this story at all. One camp instantly went full eye-roll, with the loudest vibe being “rules for thee but not for me”—a jab at U.S. companies that trained on huge chunks of the internet and are now furious someone else may have done something similar to them. Another commenter flatly called the claim flimsy, pointing to past government statements they say never held up. And then came the nerd-drama twist: one user questioned the timeline entirely, asking how Moonshot could have copied Fable only a month after release, and suggesting the report may have mixed up one Anthropic model with another.

There was even a tiny etiquette subplot when a SpongeBob-style meme got swatted down with a stern Hacker News rules reminder. So yes, the allegation is serious—but the comment section was serving skepticism, hypocrisy charges, timeline detective work, and meme-policing all at once.