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

Claude Code is steganographically marking requests

thereallo.dev

Post reply on HN
Print view View original

411–420 of 817 posts

Re: Claude Code is steganographically marking requests

#411

Post by tedivm »

next_xibalba wrote: post #302
cowl wrote: post #295

Earlier quoted context omitted.

oh no, the company that illegally used every possible media they could get their hands on is crying that some other company is doing something potentially shady but not illegal? And using that excuse to put in place hidden surveillance systems on their customers?

People keep throwing this idea around haphazardly, but U.S. courts have pretty consistently decided that training on copyrighted works falls under fair use. You may not like it, but that doesn't make it "illegal".

The courts have never said piracy, which is how the training sets were originally built, is legal. There are several court cases still ongoing over this.

Re: Claude Code is steganographically marking requests

#412

Post by 0xbadcafebee »

jfreds wrote: post #332
civet_java wrote: post #238

There are some commentors in this thread downplaying the severity of a service provider being less than transparent about exactly what their shipped tooling does on customer's machines. That the provider's business needs necessitate the this behaviour doesn't justify their lack of honest disclosure. That honest disclosure would render the solution to their problem useless isn't my problem. If anything, that they thou…

So they’re watermarking requests according to your environment variables and maybe changing a string format if you’re in a certain time zone? Am I missing something here? Where’s the five alarm fire?

No fire, just people looking for a reason to be upset. "They didn't tell us they were secretly checking for ToS violations" is their reason this time.

Re: Claude Code is steganographically marking requests

#413

Post by skissane »

danlitt wrote: post #333
zaphirplane wrote: post #317

Earlier quoted context omitted.

> Also, the output of the LLM is public domain by law Why so? Also there is a lot of code in ironically claude and ChatGPT that’s generated by LLM . Yet I haven’t seen the public domain code

The code is not eligible for copyright. If they do not give you a copy of the source code, that does not matter. And if you don't know which parts were generated by LLM, you can't safely reuse the code.

> And if you don't know which parts were generated by LLM, you can't safely reuse the code.

I speculate this could be a real issue in future copyright infringement lawsuits.

The plaintiff bears the burden of proving that the code they claim is copyrighted by them actually is copyright. If it is known that large parts of it were generated by LLM, they’d need evidence to demonstrate sufficient human input to establish copyrightability. If they’ve kept highly detailed traces of the development process, that could be rather straightforward; if they haven’t, it could be really difficult.

Now, that’s true in the US, which never accepted mere “sweat of the brow” as a basis for copyright; the UK courts have, and most of the Anglosphere follows the UK on this more than the US.

The other factor: when dealing with an (almost) trillion dollar corporation, even if you’ll win the legal argument, they may bankrupt you with legal fees before the argument is ever properly heard.

But I suspect the precedents on this topic are going to be established by lawsuits involving far smaller actors.

(IANAL and I speculate only for myself, not any present, past or future employers.)

Re: Claude Code is steganographically marking requests

#414

Post by Uptrenda »

"I think this could have been explicit. Developer tools can enforce terms. API providers can detect abuse. Companies can protect their models."

Literally, how. How does one determine what abusive use looks like for the API without context into the client? All requests look like the same stuff. If there was a better way then they would have done it. Or is the author hoping that if Anthropic writes "hey china, please don't steal our models, kthanks" they won't? Like get real. This stuff means nothing in China. China can't even manage to regulate their building industry enough to use real concrete where it's warranted.

Re: Claude Code is steganographically marking requests

#415

Post by solenoid0937 »

jameslk wrote: post #277

That's wild. If Anthropic is willing to risk ruining the trust of their userbase for the sake of protecting their moat, it makes me wonder how strong of a moat they have to begin with

No business loses trust for this, this is just standard client side anti tamper/anti RE stuff

It's a total non issue unless you're a Chinese distillation lab

Re: Claude Code is steganographically marking requests

#416

Post by solenoid0937 »

MangoCoffee wrote: post #107

The AI race right now is in a sad state. Chinese's playbook is releases open weight models and trains them on their own chips. Anthropic pushes fear and control. But the only way to win is by innovating. China is flooding the market with cheap, good enough models, while the U.S. is building a Chinese firewall.

They're trying to prevent China from reaching superintelligence, which is totally understandable when you consider the fact that the Chinese government will gladly turn its citizens into a pulp for criticizing it, censors most media to maintain absolute power, and has systemically tortured, raped, murdered, and/or disappeared most of its dissidents and human rights lawyers.

Re: Claude Code is steganographically marking requests

#417

Post by isatty »

You can't trust any of the big AI labs as far as you can throw them, and most definitely not Anthropic. They may have a good model, but they've shown time and time again that they're not trustworthy. The CEO has recently started taking a stance against local AI. That must tell you something: local AI is the future. If you want to preserve privacy and be ready for the rug pull, you need to run things locally. Unfortunately, that means that you're going to need Google or the Chinese labs to constantly release open models.

If anything, I'll trust Google more than any of the other labs just because the infrastructure that stores and protects user data was built over decades ago pre-AI craze.

Re: Claude Code is steganographically marking requests

#418

Post by cindyllm »

solenoid0937 wrote: post #416
MangoCoffee wrote: post #107

The AI race right now is in a sad state. Chinese's playbook is releases open weight models and trains them on their own chips. Anthropic pushes fear and control. But the only way to win is by innovating. China is flooding the market with cheap, good enough models, while the U.S. is building a Chinese firewall.

They're trying to prevent China from reaching superintelligence, which is totally understandable when you consider the fact that the Chinese government will gladly turn its citizens into a pulp for criticizing it, censors most media to maintain absolute power, and has systemically tortured, raped, murdered, and/or disappeared most of its dissidents and human rights lawyers.

[dead]

Re: Claude Code is steganographically marking requests

#419

Post by solenoid0937 »

applfanboysbgon wrote: post #281
SubiculumCode wrote: post #82

Earlier quoted context omitted.

1st, this technique is not fraud, and fraud is a separate accusation. 2nd, paying customers can legally and legitimately be banned and monitored for breaking terms of service, which probably includes things like using the model against U.S. export restrictions.

> 2nd, paying customers can legally and legitimately be banned and monitored for breaking terms of service Yes, I said that. If a user is breaking your terms of service, ban them. Continuing to charge them while not providing the service they're paying for is, in fact, literal textbook fraud.

No, it's totally legal to provide degraded quality of service to those breaking your ToS.

In any case this is not what is happening, but it is legal.

Re: Claude Code is steganographically marking requests

#420

Post by BeetleB »

gunapologist99 wrote: post #383
NewsaHackO wrote: post #296

Earlier quoted context omitted.

Half of those don't actually require proxying Claude. Also, Claude has made it apparent time and time again that it does not want people using Claude Code as a "tool" in a workflow. If you want to select a model dynamically based on the prompt difficulty, Anthropic wants people to use the API for this. It was the whole issue Claude had with OpenClaw.

> Also, Claude has made it apparent time and time again that it does not want people using Claude Code as a "tool" in a workflow. Why would Anthropic get to dictate how someone uses a "tool" (that's literally what Claude Code is... a tool in a workflow) They're swimming upstream. Trying to maintain a rapidly shrinking moat and not being very creative about it. Making enemies of your users is often a failing strategy.

[flagged]
Post reply on HN