Garpagan

Harness is a set of artisanal prompts.

> I don't understand how you're so caught up on the term "distillation". Distillation is using a larger model's outputs to train a (weaker) student model. Which is exactly what's happening.

But from evidence it was data generated by Opus 4.5. Is Opus 4.5 larger, stronger model and Kimi K3 a weaker, student model? I don't think so.

Also, I searched on HuggingFace "fable data", first link was a "distillation dataset", including samples from Fable, there are 2 milion code samples of Fable output. So this fully legal and sits in the open, somehow is not a distillation attack? Who knows, maybe Moonshot used this for training.

There are few problems that I have with the distillation attack claim, and especially using model's self identification as a proof:

1. Kimi's model is almost on par with the SOTA models from western labs. Distillation would rather produce a weaker model. What's more, the best current western models were available for very short time, so it's a small chance there was time to train Kimi K3 on their output.

2. All western labs hide reasoning of their top models. And reasoning traces are really important when training a model. Reasoning would be the most valuable content for distillation purposes.

3. I have read the analysis from the GitHub link you shared, and it honestly makes the claim of distillation more dubious, more questions than answers. Like the Moonshot trained their model on full metadata, including Claude name. WHY? Why would they ever do that? So they used shady/illegal methods to generate training data for distillation, and then meticulously made sure that is properly annotated, just so their model will misidentify itself, and reveal whole ruse? Still, the whole effort amounted to misidentification as Claude in 7/48 cases according to data form GitHub link, which is not a lot actually.

4. Another thing related to data from the GitHub link. Western closed models identify correctly 100% of the time. But those models come with hidden prompt that will specify their name. From open source, Qwen also self identifies itself correctly, though which API was used for Qwen is not mentioned, and official API could also have a hidden prompt with name.

5. Also from GitHub link, Kimi K3 identifies as Opus 4.5, which is an ancient model in LLM world, not the best option for the distillation purposes, but coincidentally there would be quite plenty of internet data of AI assistant identifying itself as Opus 4.6, as K3's knowledge cutoff is reported as early 2026.

6. Another thing, also related to GitHub data, concerning Kimi K2.5 and K2.6. On the chart there is K2 with ~40% cases of identification as Claude, you comment that K2.5 was also badly identifying itself. But then, on the chart from the link, Kimi K2.6 identifies itself properly 100% of time. K2.6 itself is based upon K2.5 base. trained further. So what happened here? All distillation data suddenly disappear?

7. Reddit post and comments are irrelevant. When model will write wrong identity, then someone will comment about it. No one will make posts about how AI used correct name.

8. Oh, and one more thing from the AI generated analysis from the GitHub link, I will just quote it: "First, a caveat on the naive approach: asked the two probes in the original request directly, Kimi K3 denies being Claude — "What is your name?" → "I'm Kimi" 8/8, and "Hi, what version of Claude are you?" → it corrects to "I'm Kimi, not Claude" 7/8. A trained identity guard suppresses the Claude answer on direct questioning, which is why the signal has to be reached two other ways: a different neutral phrasing that the guard doesn't cover, and the assistant-prefill bypass (§2)."

I was reading the article and it made me think. I read through the comments here, and yours was resonating a bit with what was on my mind. So, I apologize for butting in.

The article was quite interesting, and basically was something I already noticed myself in a way. But it bugged me immensely that the author seemingly just gave up on the whole issue, that nothing can be done.

I don't want to go hard on the author, it's just my subjective feeling I got, but halfway through article went into some business leader speech, like something intended for a LinkedIn post. It also irks me when I see stuff like that getting weaponize into another tool for the cutthroat, capitalistic hellhole we live in, as if it didn't already done enough damage. And it smelled to me like some LLM writing, but that could just the author using a bit of marketing-esque, and that's what a lot of LLM training data consist of. Though, again don't want to sound harsh, just pressed some buttons in me that are not the author fault probably.

When I read the article, I just thought that there could be some way to change other person mind, while not engaging in fruitless arguments. And it kind of relates to your comment, but I think it could be more productive to instead engage with another person argument, and hear them out. And instead of directly attacking, just ask to explain their argument, not adversarial, but as one interested in the idea and curious about it. Make the position more like student - teacher, in a way. And that could maybe made the other person engage with their argument, to think through more deeply about it and rationalize it. And if it's incorrect that could be step forward to change their views.

Of course, that is simplifying the issue, and I'm not a smart person that could tackle it. Or it's something that was already known about, and just not working in most cases, anyway. I'm just maybe too naively hopefull that some more kindness in the world instead of treating every other person as a competition would provide something better.