Several top AI researchers have already chimed in and said this is not technically or physically possible.
Senior White House official claims China’s K3 model stolen from Anthropic
Donald Trump’s Assistant for Science and Technology, Michael Kratsios, has accused China’s Moonshot AI of creating its head-turning Kimi K3 model distilling Anthropic’s Fable. Kimi K3 is 2.8-trillion-parameter open-weights model of such impressive quality that its mere existence suggests Chinese AI researchers aren’t far behind …
COMMENTS
-
-
Thursday 23rd July 2026 11:28 GMT Charlie Clark
Re: US Treasury Secretary Scott Bessent...
Part of the "Donroe doctrine", apparently:
Scott Bessent, the treasury secretary, recently described as “economic statecraft in the 21st century”, in which global access to the dollar and the American economy is “no longer unconditional”
From The Economist last week: the rules-based order is over, you do what you're told or we'll cut you out.
Personally, I'd call this economic and diplomatic lunacy by treating friends and allies as potential adversaries…
-
-
Thursday 23rd July 2026 06:36 GMT amanfromMars 1
Another 0day Vulnerability Exploited, Another Tantrum Delivered
Do you not despair, Simon Sharwood, El Reg APAC Editor, at reporting such predictable events as if new news rather than treating them like they really are ... rotten dishes of green cheese and sour grapes ..... whenever one cannot fail to recognise the stench of it surrounding them‽
Senior White House officials make ab fab fools of the USA again with claims China’s K3 model stolen from Anthropic.
Donald Trump’s Assistant for Science and Technology, Michael Kratsios, and US Treasury Secretary, Scott Bessent, throw their toys out of the pram.
-
Thursday 23rd July 2026 06:42 GMT EricM
"Distillation" is just training on Question-Answer pairs...
The questions come from the entity doing the distillation. The answers are AI-generated, so they are not even copyrightable. The distill-training with this information is done by the entity that formulated the questions.
Wonder where they construct "stealing" here...
What exactly was "stolen"?
-
-
Thursday 23rd July 2026 08:34 GMT Anonymous Coward
Re: "Distillation" is just training on Question-Answer pairs...
So digging that out and sharing it is a bad thing thing? Shouldn't you be encouraging everyone, big and small, to join in this "distillation" from all of the big LLMs and giving back (the munged results from) what was stolen?
Which leaves the open issues that this CLAIMED distillation occurrence was by another big company, who won't be sharing, and one that happens not to be in the US.
-
-
-
Thursday 23rd July 2026 07:56 GMT EricM
Re: "Distillation" is just training on Question-Answer pairs...
Yeah, but that part is supposed to be "legal" as per the AI company's arguments, based on the assumption that training is different from copying...
Which, curiously, adds another question mark to this "distillation equals stealing" statement, as distillation is just another form of training.
-
Thursday 23rd July 2026 08:28 GMT Anonymous Coward
Re: "Distillation" is just training on Question-Answer pairs...
That is not the point of the question you are responding to, as well you know.
The appropriation of the training data by the US AI companies is one matter to be addressed. Correct.
But the point here is that the same US company that committed the act you (correctly) dislike is being used as a pawn by the White House official who is claiming that a different act of theft has occured. But by the "rules" that are sort-of being applied to the use of the US AIs this claimed "distillation" is perfectly ok.
-
-
-
Thursday 23rd July 2026 09:40 GMT Bebu sa Ware
We don't like it…
and we hate even worse there is nothing we can do about it. (As in "The US has already sanctioned just about every major player in China’s tech ecosystem")
One can only hope that this is purely theatre because if the US Administration does not appreciate the technological progress of the PRC ironically kicked into high gear by Trump's "policies", they will find that an unpleasant surprise awaits.
-
Thursday 23rd July 2026 12:00 GMT DrXym
Photocopying
I'm sure you can "distill" knowledge from one AI to another if you know what to ask but I bet the distillation feels a bit blurry and vague like photocopy. It conveys the same information but it has captured errors, as if creases, paper clips or your finger were captured in the copy.
-
Thursday 23rd July 2026 13:42 GMT Groo The Wanderer - A Canuck
Based on no presented evidence, proof, or suggestions that said will ever be provided.
"We will NOT tolerate competition with our billionaire overlords and Pumpkin Fuhrer in the US of A, especially not from open source that doesn't even steal our revenue!"
-- White House Official Arse Kisser
-
Thursday 23rd July 2026 15:28 GMT 2141
Decline and Fall
China is demonstrating it's imminent dominance in AI and robotics, and it is leading the moment with proposals for a global regulatory framework.
The USA is on the back foot, whining about a clever nimble adversary. No global ideas. America's answer is to take its ball home in a sulk. A vast strategic error when China is offering a ball to everyone!
-
Monday 27th July 2026 09:07 GMT rgjnk
Bollocks
Throw a little bit of critical thinking at it and the claims fall apart.
Does anyone seriously believe that they could have done it - even if technically possible - given the tiny amount of time thats passed, the cost of access, and the limits on access, and that even if they somehow could that this wouldn't have been wildly obvious very early on?
I quite believe that we're talking about people who could and would steal anything to get an advantage but I suspect the truth is that they didn't, and that it's this that really scares those in the US.
It's not like there's any magic involved here unique to the US, it's clever stuff but the only real barriers are data, compute and money and all those are available to anyone motivated enough.