Claude Code is steganographically marking requests • Hacker News Board

Re: Claude Code is steganographically marking requests

#31

Post by dofm » Tue, Jun 30, 2026, 4:15 PM UTC

atonse wrote: post #14

[flagged]

Not totally new territory; there was a highly compressed period of panic about encryption 35 [0] years ago:

https://en.wikipedia.org/wiki/Pretty_Good_Privacy#Criminal_i...

[0] f**k I'm old

Re: Claude Code is steganographically marking requests

#32

Post by sigmoid10 » Tue, Jun 30, 2026, 4:16 PM UTC

If they only collect the data for analysis I guess this is fine (they already get way more sensitive data from users anyways, so if privacy is your concern you've made the mistake many steps ago). The much more interesting question is if they directly act on this data in their API. For example by rate-limiting, compute-limiting or rerouting to weaker models. That might even be legally questionable. I would really like to see this as a follow-up analysis, but I guess it is way more difficult and will also cost quite a bit in tokens.

Re: Claude Code is steganographically marking requests

#33

Post by love0972 » Tue, Jun 30, 2026, 4:16 PM UTC

Is that really how it is? How will this affect our future?

Re: Claude Code is steganographically marking requests

#34

Post by dylan604 » Tue, Jun 30, 2026, 4:16 PM UTC

dwa3592 wrote: post #25

saddlerustle wrote: post #15

[flagged]

this seems a bit extreme. pangram does not work. i have tricked it multiple times. i don't get how people are still trusting these systems.

it's just a different car on the hype train

Re: Claude Code is steganographically marking requests

#35

Post by throwawayffffas » Tue, Jun 30, 2026, 4:17 PM UTC

Claude code does feel very malwarey to be honest. They have been like that from the start.

Re: Claude Code is steganographically marking requests

#36

Post by wolttam » Tue, Jun 30, 2026, 4:18 PM UTC

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.

Re: Claude Code is steganographically marking requests

#37

Post by Maken » Tue, Jun 30, 2026, 4:19 PM UTC

atonse wrote: post #14

[flagged]

If scrapping content is legal, model distillation should be legal too.

Re: Claude Code is steganographically marking requests

#38

Post by dylan604 » Tue, Jun 30, 2026, 4:19 PM UTC

axutio wrote: post #30

midtake wrote: post #13

[flagged]

Would you also say that "someone who wants to use an IDE / LSP features to code and not give credit to the IDE / LSP is the worst kind of person"? If not, what is the difference between the two for you?

one wrote code while the other is used by meatbags to write code. why is this example always marched out like it means something?

Re: Claude Code is steganographically marking requests

#39

Post by Klonoar » Tue, Jun 30, 2026, 4:19 PM UTC

If there weren't already enough tells that something is AI-generated, I guess you could add this to the list.

Re: Claude Code is steganographically marking requests

#40

Post by kej » Tue, Jun 30, 2026, 4:20 PM UTC

gonzalohm wrote: post #21

100ms wrote: post #11

What's the point of even trying to obfuscate this with such a simple method? Could at least have hidden the targeted features by storing their hashes or embedding a bloom filter or similar

The point is not raising red flags I guess

I love how well this comment works as a vexillology joke, even if it wasn't intended.