Live data from Hacker News

Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

reuters.com

291–299 of 299 posts

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#291

All remote AI are a massive security risk for individuals/companies/governments that may be targeted by the US government. It is likely that the US will get a live feed from each AI provider that they are inspecting in real time to identity things of interest, terrorist attacks or foreign government planning or even foreign companies competitive to key US companies. It will give them access to the though process in t…

Why make this u.s. centric? You think China served models would be different?

Because Chinese model providers are open source. You can host them yourself.

Can you name one open source US model that came out in the last year?

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#292

Earlier quoted context omitted.

It is worth thinking about the fact the total throughput of even a big LLM provider isn't many megabits. If a token compresses to around a byte, worldwide AI input and output is around 1 gigabyte per second. For any intelligence agency, they can afford to keep and store all of that forever, and later do analysis on it.

> For any intelligence agency, they can afford to keep and store all of that forever, and later do analysis on it. At the scale the AI companies are operating at, I think it isn't likely that they are sucking it all in right now. More likely I think the intelligence agencies will get a real-time live tap into the raw data feed which they will process onsite for interesting things and then if things are flagged, they…

But for just a gigabyte per second, why not just take and store it all?

The cost to do so for many years is less than one employee.

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#293
post #290

Earlier quoted context omitted.

Are you still talking about China? Where they are denied food, shelter, healthcare? Or do you accept the CCPs word over the word of the people who show videos of what reality is like in China?

I don't need to accept anyone's word, I've traveled both the US and China extensively. I'm noticing you're trying really hard to deflect though.

I can tell you haven’t. I lived in China for 8 years.

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#294

Earlier quoted context omitted.

Why make this u.s. centric? You think China served models would be different?

Because Chinese model providers are open source. You can host them yourself. Can you name one open source US model that came out in the last year?

Gemma 4 just came out 2 months ago

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#295
being true or not is irrelevant for decisions such as this. has been done on both sides, whether at software level or "hardware".

from Teslas not allowed parked around sensitive areas in the city, to blocking a (very famous and quite well-made) Russian antivirus, or Huawei communication stack.

Anthropic is not doing itself any favours with their recent (?) antics [1], so it is completely well-founded to do this imho. regardless, harness as a moat is not quite as established as the underlying models to the same extent.

[1] https://news.ycombinator.com/item?id=48734373

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#296
post #250
post #246

Earlier quoted context omitted.

> The Great Leap Forward wasn't "killing" people, which implies intent. It was just good old economic mismanagement. If both the USSR and the CCP had millions killed in the process of modernization, without stopping when knowing the death toll, maybe there's intent after all? How would you describe the cultural revolution then? another case of economic mismanagement?

I noticed you haven't addressed my main point at all. What are the millions dying of poverty every few years in the US (in a country with like a quarter of the population!), a death toll that still hasn't been stopped ? Is there intent there as well?

That study statistics seem questionable at best just so to say the US should have universal health care. A lack of policy is not the same as enacting a policy, especially one that is supposed to emulate past USSR actions which they knew at the time led to millions of dead. This makes it hard to say the CCP didn't know what would happen.

Later when it was found out mass starvation to the point of wide spread cannibalism was happening they would not stop, yet continue sending their troops to steal the peasant's food and execute anyone who resisted.

We also of course have Mao quoted as saying "Death has benefits; fertilizer is created", "Half of China may well have to die", as many other quotes showing they weren't really interested in the millions that will die.

This culminating in 30 million dead in 4 years, which if we take your study at face value will take the US 100 years to achieve through "poverty related deaths", which even communist utopias have.

Regarding the Cultural Revolution I assume you agree this was intentional murder of a million+, although you seemed to challenge that in my original posting. Our current discussion seem to focus on intent even though I said "killing" which usually does imply reduced intent.

Regarding the Great Leap Forward, I do believe the CCP would prefer no deaths, and a lot of the deaths relate to strongman idiotic ideas without challenge, as is endemic in ideological totalitarian societies. However, I believe they fully knew what they were going into, without taking into account the refusal to stop, and therefore these were horrendous crimes.

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#297

Earlier quoted context omitted.

Why make this u.s. centric? You think China served models would be different?

Because the topic of the article is about the US?

No, the topic was about Alibaba and Anthropic.

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#298
post #240

Earlier quoted context omitted.

I wish people would stop using Anthropics incorrect use of the term distill. They don’t share logits so you can’t distill. You can generate training data, which doesn’t sound nearly so scary.

Why do you need logits to distill? Those are at least tokenizer-dependent, and different models use different tokenizers.

Distillation is a specific named technique from a pretty famous paper "Distilling the Knowledge in a Neural Network," by Hinton and others.

What makes it different from just training on any data is that you train the student model on "soft targets" that includes the full output distribution (logits) from the teacher model. Regular training uses one-hot targets and penalizes anything else; distillation will partly reward a student for getting in the distribution. This teaches the student to think like the teacher, not just imitate it.

Generating training data is not distillation, not in the technical sense, and I dislike Anthropic undermining the nomenclature to set a narrative.

Re: Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says

#299
post #298

Earlier quoted context omitted.

Why do you need logits to distill? Those are at least tokenizer-dependent, and different models use different tokenizers.

Distillation is a specific named technique from a pretty famous paper "Distilling the Knowledge in a Neural Network," by Hinton and others. What makes it different from just training on any data is that you train the student model on "soft targets" that includes the full output distribution (logits) from the teacher model. Regular training uses one-hot targets and penalizes anything else; distillation will partly rew…

Thanks, I wasn't aware of that paper!

I think the popular use of distill (one-hot rather than logits) is not due to Anthropic but due to the DeepSeek-R1 tech report: https://arxiv.org/abs/2501.12948

Post reply on HN