Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Anthropic says Alibaba illicitly extracted Claude AI model capabilities

reuters.com

Post reply on HN
Print view View original

701–710 of 1001 posts

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#701

Post by Upvoter33 »

itopaloglu83 wrote: post #614
breput wrote: post #16

Earlier quoted context omitted.

I think you meant a quote attributed to Bill Gates: "Well, Steve, I think there's more than one way of looking at it. I think it's more like we both had this rich neighbor named Xerox and I broke into his house to steal the TV set and found out that you had already stolen it."

I thought Xerox demoed something they haven’t implemented yet, and Apple turned a mockup into a real GUI.

No that is not true. Read about PARC and all the crazy tech they built some time. It was ahead of its time!

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#702

Post by xela79 »

I would say Antrophic and others illicitly extracted free internet content and put it behind a paywall, giving zero compensation to those that made their whole business possible in the first place. So smallest violin player busy here trying to make me care if it happens to them.

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#703

Post by Laurel1234 »

dist-epoch wrote: post #601
anon373839 wrote: post #380

Earlier quoted context omitted.

> These complaints of distillation are inflating the problem to make it sound worse than it is Unfortunately, the Reuters piece itself is complicit in this dramatization. The lede paragraph parrots Anthropic's talking point that distillation is an "attack", without using quotes that would alert the reader that this framing is a corporate talking point. Distillation is NOT an attack.

Reuters is probably the most rigorous news agency in the world. > it said was the largest known attack > Anthropic said in the letter it was supportive of the U.S. government's efforts to combat the attacks both times the word "attack" appears it's clearly stated that the word was used by the company, it's a direct company quote. actually putting it into quotes would be editorializing > Unfortunately, the Reuters pie…

> how would you feel if somebody quoting you would turn your word dramatization into "dramatization" because they don't agree with your assesment

This is exactly what news agency should be doing though. When the dude showed up to Comet Pizza to look for Hillary Clinton or whatever, do you figure they should've printed "Local hero saves children from predatory cabal"?

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#704

Post by noosphr »

itopaloglu83 wrote: post #614
breput wrote: post #16

Earlier quoted context omitted.

I think you meant a quote attributed to Bill Gates: "Well, Steve, I think there's more than one way of looking at it. I think it's more like we both had this rich neighbor named Xerox and I broke into his house to steal the TV set and found out that you had already stolen it."

I thought Xerox demoed something they haven’t implemented yet, and Apple turned a mockup into a real GUI.

They changed the live system from having line by line scrolling to pixel scrolling after Jobs asked why they didn't do it during the lunch break.

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#705

Post by bryceneal »

tristanj wrote: post #681
bryceneal wrote: post #651

Earlier quoted context omitted.

I have 0 sympathy for Anthropic. Their latest models are extremely censored. The Fable rollout was horrible. Their Cyber Access program criteria denies doxxed Americans doing legitimate security work. Anthropic is hostile to their users and hostile to their own country. OpenAI is considerably better on all of these fronts, but still not perfect. I'm happy to use and support Chinese model developers if it means less c…

Chinese models are the exact opposite of what you claim to want, they are all highly censored, even more so than Anthropic models, with government mandated censorship.

Your take does not reflect the reality on the ground. The Chinese models are censored on a narrow range of political topics which have nothing to do with my work. The weights are open and they can be uncensored/abliterated with little effort.

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#706

Post by Laurel1234 »

macleginn wrote: post #627
ACCount37 wrote: post #590

Earlier quoted context omitted.

It's not about how big your dataset is - it's about how you use it. I jest, but I'm also completely serious. 1T tokens from Claude can teach a model something 1T tokens scraped from the open web can't. Things like "how an LLM can problem solve effectively", or "how an LLM should use tools", or "how to construct reasoning chains", or "when to double check", or "what innate capabilities an LLM can or can't rely on". Th…

Unremarkable base model will remain an unremarkable fine-tuned model that memorised a couple thousand of input-output pairings.

If that were the case, Anthropic wouldn't be throwing a fit over distillation "attacks".

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#708

Post by sorenjan »

0xbadcafebee wrote: post #11

There's two basic kinds of distillation: 1) the massive [and dumb] method where you ask a question and use the answer as reinforcement (Black Box), and 2) more targeted distillation where you use one model to directly inform/train/guide another model (RLAIF). The latter is basically fine-tuning the model with direction from another model. Thousands of businesses do this every day to fine-tune. This is almost certainl…

Doesn't "real" distillation use the logits instead of the final tokens? I would classify this more like using a model to generate synthetic training data.

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#709

Post by lambdaone »

xdennis wrote: post #301
protimewaster wrote: post #153

Earlier quoted context omitted.

The AI companies seem to take the viewpoint that everything on the internet is free, except their stuff. It's okay to hammer some random website with AI crawlers, ignoring robots.txt, and causing bandwidth costs to skyrocket. But if you cost an AI provider money with your data acquisition practices, well, that's just clearly unacceptable.

That's one aspect, which is a bit of a gray zone. But Anthropic trained on pirated books. That is explicitly illegal.

As I understand it what was "explicitly illegal" was copying the books, in the sense of mere copying before feeding them to the model, and this is what the Anthropic copyright settlement is about.

Actually processing them through the model, though, was considered transformative and therefore fair use.

Re: Anthropic says Alibaba illicitly extracted Claude AI model capabilities

#710

Post by Freedumbs »

This article is absurd for an outlet who published an article that's meant to be news not editorial. Reuters was once a news wire and is still considered that. The first two paragraphs refer to "attack" and "strike" against Anthropic. This is sensational nonsense, not news. There was no strike, or attack. Block the accounts. Why is this news, and why are they pandering to the people who just banned the new model they burned at least $10 billion training? The closer you look at this AI stuff the more absurd it is. I assume the strat is to keep the bubble floating until post-2028, then drop the bomb on the Dem who wins. Just like with the covid inflation + economic rigging Trump did in 2016-2020.
Post reply on HN