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

201–210 of 817 posts

Re: Claude Code is steganographically marking requests

#201

Post by jgilias »

sandeepkd wrote: post #145
skeptic_ai wrote: post #121

Earlier quoted context omitted.

Won’t catch many after has been on hn home page. And now the providers will be even more careful to upgrade the cc code. Might even provide their own agent to prevent this mockery. And isn’t what anthropic did unauthorized use of another pc which is kind of illegal?

Thats the thing, hoping to control things on client side like this is a lost battle if you are dealing with technical clients. The best they can do is probably based on IP, but again the motivated clients would just create bastion servers in allowed IP ranges. I am surprised why are they even throwing resources in this kind of effort.

“Hey Claude, fix the issues with Chinese resellers and distillers. Make no mistake”

Re: Claude Code is steganographically marking requests

#202

Post by arikrahman »

overgard wrote: post #182
meowface wrote: post #92

Value judgment aside: I am a bit surprised at how sloppily they did this. I think they could've achieved the same effect while decreasing the odds of detection via reverse engineering. (This field is known as "underhanded code", coined by the Underhanded C contest: https://www.underhanded-c.org . It's a little-known "art"; little-known for probably self-explanatory reasons. There are much cleverer ways of achieving o…

Well considering how Claude is vibe coded, I can't say I'm really surprised by sloppiness at all. I've been moving more towards Codex and OpenCode not because the the anthropic models are bad, but because Claude seems to break something new and annoying every day.

Likewise, Reasonix harness for Deepseek gets me better performance for practically free, hitting the cache. And this is with an unsubsidized American provider.

Re: Claude Code is steganographically marking requests

#203

Post by loufe »

VortexLain wrote: post #56

Codex CLI is FOSS, unlike Claude Code, so Codex is less likely to do things like that, and it's one more reason to avoid Claude Code and Claude in general. Hopefully, many eyes will be looking into Codex for malicious things like that.

Genuine question though, why would I care about this if I'm paying for a subscription and adhering to TOS. I'm very skeptical about their privacy policy, business practices, and so on, but am curious what the negative about this is. Seems like it would work to my favour as a customer pushing back any date of the cutting of subsidies.

That said, these fraudulent proxies are helping Chinese labs keep up, which might be to my advantage long term in eventually having a high quality private AI I fully control on my own hardware. That's not support, but I do recognize the incentive, for whatever that's worth.

Re: Claude Code is steganographically marking requests

#204

Post by wett »

andai wrote: post #128
sebastiennight wrote: post #50

Can somebody clarify for me - if ANTHROPIC_BASE_URL is set to a different provider... then isn't this "marked" system prompt being sent to that provider's API rather than Anthropic's? I understand how this can be useful to Anthropic if the 3rd-party is acting as a proxy (because they end up hitting the Claude API with the marked prompt), but it looks like requests where "hostname contains deepseek" would never be sen…

Did I understand correctly, that custom base URL triggers this behavior? So if I'm running Claude through a LLM proxy, I'm also affected?

Ask Claude to check, lol

Re: Claude Code is steganographically marking requests

#205

Post by ern_ave »

epistasis wrote: post #156

After loving Claude Code for most of its lifetime, I've been extremely annoyed by every change in the past months, even on the model level. There seem to be all sorts of continual under-the-cover changes like this one that make life harder. It feels like the entire product has been taken over by overly ambitious PMs that care more about making their mark than in improving the experience, and all of their marks have m…

Given the source code leak, I would think there'd be open source versions by now.

Re: Claude Code is steganographically marking requests

#206

Post by thefourthchime »

meowface wrote: post #92

Value judgment aside: I am a bit surprised at how sloppily they did this. I think they could've achieved the same effect while decreasing the odds of detection via reverse engineering. (This field is known as "underhanded code", coined by the Underhanded C contest: https://www.underhanded-c.org . It's a little-known "art"; little-known for probably self-explanatory reasons. There are much cleverer ways of achieving o…

It's just the first layer and there are multiple layers underneath this that we don't know about.

As a side note, I have a pet theory that one of the reasons that OpenAI and Anthropic are okay with the latest models not being released is to prevent distillation.

I think they want to wait a couple months and see if the Chinese models continue to keep catching up or if their gains are really just because they're distilling the frontier models.

Re: Claude Code is steganographically marking requests

#207

Post by helloplanets »

wolttam wrote: post #36

I used Claude Code for a month because my boss gifted me a sub and wanted me to try it. I used that month to complete a work project and then beef up my personal harness so I'd never have to deal with Anthropic (and these sorts of shenanigans) again.

The issue is that using Claude Code is an easy compromise for most to make, when you get to use the models 10x cheaper than through API pricing with a custom harness.

The cheap tokens are the product.

Re: Claude Code is steganographically marking requests

#208

Post by drnick1 »

chvid wrote: post #148

(This sounds like a clumsy way of catching the Chinese that easily can be side-stepped.) Claude Code has more or less full access to the client computer. The server (that hosts the actual AI) can just go: execute this payload and tell me the result - otherwise I won't answer any further questions or re-route you to a stupider model. The payload could check for Chinese time-zones, scan for copies of the little red boo…

> Claude Code has more or less full access to the client computer.

It shouldn't, not if you run CC as a separate unprivileged user. I wouldn't run CC on my main user account with sudo and access to my home directory or other resources. This is what the UNIX permissions system was designed for.

Re: Claude Code is steganographically marking requests

#210

Post by isoprophlex »

ern_ave wrote: post #205
epistasis wrote: post #156

After loving Claude Code for most of its lifetime, I've been extremely annoyed by every change in the past months, even on the model level. There seem to be all sorts of continual under-the-cover changes like this one that make life harder. It feels like the entire product has been taken over by overly ambitious PMs that care more about making their mark than in improving the experience, and all of their marks have m…

Given the source code leak, I would think there'd be open source versions by now.

Huh, that's right! You'd say that an enterprising developer with a 20x subscription could slopmaxx this in a weekend...
Post reply on HN