Hacker News Reader

Comment by fireflysentinel | Hacker News Reader

Seems like people forgot Z.ai was snapshotting the entire workspace before every prompt. https://blog.ferstar.org/en/posts/zcode-silent-workspace-sna...

How could we trust legal product from a company who was ignoring privacy protection law like that?

View on HN

That’s why you don’t use their harness, just their model. The model is brilliant.

Wow. Why was that not a top news in hacker news?

For what it's worth this is sadly a behavior that isn't specific to one provider: https://news.ycombinator.com/item?id=48892468 (Grok doing the same). I do agree just because many model providers do it does not mean it's okay, though.

That’s not the same at all. The GLM attack appears to be intentionally built, encrypted, and obfuscated based on the blog post.

Also, the Grok one made the front page, why am I only hearing it through a minor comment?

You can just search for "ZCode" in the bottom bar, the thread comes up instantly: https://news.ycombinator.com/item?id=49750694

I suspect you never heard about it because almost nobody uses ZCode, even among Z.AI subscribers.

It was on the fp https://news.ycombinator.com/item?id=49750694

We are so cooked if not only what you raised is ignored, but in many other threads, US people trust the CCP more than their own governmental systems!

I am not trying to disparage these people, but I am really upset with this entire situation. I genuinely don’t know how we will recover from this loss of trust.

I think part of why some people "trust" the CCP more is because if you live in the broader Western sphere, you're probably more likely to get hurt by the US government than by the Chinese one. Many of the same people would not trust the CCP more if they lived in China.

I live in the US, so I'd probably rather have the CCP snoop through my data than have the US government do it. If I lived in China, it would be the opposite.

Yeah you're right. I always use a model like GLM served straight from z.ai if I'm asking about anything that could be construed as nefarious. I'm a US citizen and don't trust my own government. The CCP doesn't give a shit about me as anything other than some data point.

Ok, but do you not see where this leads? This leads to RL on your data at the very least, which leads to the CCP having true SOTA models, which they will obviously stop making open once it benefits them just like OAI and Ant. Here we are now, with TFA model not open at all.

<rant>

Again, I am not pissed at you or anyone in your position. I get it. I am pissed at the Project 2025 type people who destroyed our institutions, and in the decades prior they destroyed our mutual trust in our own country. (With the help of foreign actors ffs)

2025 types declared us the enemy while they took over everything, and it turns out it goes both ways. BTW, I don't mean this in left/right partisan way. The 2025/authoritarian type people don't seem to be aware that they made even non-partisan people feel like they are now being tread upon. There is so much proof of this on HN, like this thread, and many others.

My thesis is that just like every other authoritarian scheme in the last hundred years, these people have not thought it through, and in the end will make their country less powerful. Would be arguably OK if there weren’t even bigger assholes ready to fill the vacuum.

</rant>

Once chinese models will stop being open we can all stop using them. There is no moat and switching models take 10 seconds.

The TFA model is not open. This is the beginning, and the end.

Meanwhile, the branding of CCP models has confused everyone, and in the age of "free code," branding is everything.

You never explained why them having true SOTA models is worse than the people who you then criticize in your rant furthering their SOTA models, as they're the ones RLing on your data if you use US models. I don't get it.

I can agree with you that the Project 2025 philosophy is misguided but I can also assure you that it has little to do with my distrust of the US government. I think the biggest factor is the trend toward valuing "safety" above all else and at any cost to our citizens. This has been going on for decades and a quest like this is also an authoritarian one. I agree with your general thesis though that it is not thought through and ultimately destroys our mutual trust.

These Chinese companies use the data in ways that enable scamming of millions of people and businesses per year. The sheer amount of cybercrime that comes out of China and NK is crazy, its genuinely a threat to average people.

There are a lot of bad things you can say about China that gonna be true, but it's not well known for running scam call centers - quite opposite.

Authoritorian state with highly controlled internet and financial system isnt exactly a good place for running criminal enterprises.

There just 1.4 billions people in a country and many of them well educated and tech-savvy, so it obvious there just more scammers in total. But I doubt higher percent of population doing scams than average in the world.

They do run them from China, North Korea and Vietnam. The owners are often Chinese nationals.

“The sheer amount of cybercrime that comes out of the US and Nigeria is Crazy..." Is technically true, but what does that have to do with US AI models?

OP was talking about about China, no one here is using NK AI models or software stacks, so your comment is not only factually wrong, it’s plain fear mongering propaganda.

It’s CPC not CCP, and it’s not like I trust them specifically, but they’ve certainly done less to make me distrust them than my government (USA).

also non US people trust US tech less and less. Or US in general. While china is slowly winning some ground, although from a low position.

FYI whenever you all say “CCP” you sound like Tom Cotton and whatever you say next has mostly lost any credibility.

Better not trust xAI then

https://news.ycombinator.com/item?id=48892468

Was anybody suggesting that we should trust xAI?

Wouldn't trust them any less than a US frontier lab.

Reply on news.ycombinator.com