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

Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

github.com

Post reply on HN
Print view View original

211–220 of 282 posts

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#211

Post by hurril »

the_af wrote: post #205
hurril wrote: post #184

Earlier quoted context omitted.

This is again conflating at least two things and this is so prevalent in this context. Let us not conflate how annoying DRM:s are to us users that buy the things, with pirates thinking they somehow have a right to use any software without paying fairly for it. I would even go as far as to say that you pirates are the reason I have to have a DRM in the shit I bought and paid for.

> I would even go as far as to say that you pirates are the reason I have to have a DRM I think this is largely an incorrect take. DRM is anti consumer, not anti piracy. In fact, it has done very little to deter actual piracy (and remember it only takes ONE person to break the DRM), while affecting some casual pirates and all legitimate users. In the process, they got rid of reselling stuff you own. It's anticonsumer…

Would the DRM exist without piracy?

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#212

Post by windsurfer »

swingboy wrote: post #194
windsurfer wrote: post #192

Earlier quoted context omitted.

Even if you remove a watermark, the companies still have a record of which images they have generated and for whom. Even if you remove the obvious watermarks, all major image generators are using steganography to embed hidden information that you can't be sure were removed. This is a type of one-sided arms race where one player gets to be invisible if they want to.

Do they still work if you apply something like a filter or additional layers on top? Or add a subtle blur, etc.

Yes usually, since an important aspect of steganography is error correction. For example we know that SynthID is robust enough to survive resizing and small blurs.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#213

Post by jimmydorry »

hurril wrote: post #211
the_af wrote: post #205

Earlier quoted context omitted.

> I would even go as far as to say that you pirates are the reason I have to have a DRM I think this is largely an incorrect take. DRM is anti consumer, not anti piracy. In fact, it has done very little to deter actual piracy (and remember it only takes ONE person to break the DRM), while affecting some casual pirates and all legitimate users. In the process, they got rid of reselling stuff you own. It's anticonsumer…

Would the DRM exist without piracy?

They succesfully did away with 2nd hand markets and the concept of "owning" anything. So yes, I would imagine DRM would continue to exist without piracy.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#214

Post by Planktonne »

b65e8bee43c2ed0 wrote: post #196
Planktonne wrote: post #157

I find the casual malevolence of this kind of thing breathtaking. Every time someone raises potential issues with AI misuse, they're dismissed as fear-mongering while a host of people immediately rush to demonstrate that not only is the fear justified, it's gleefully anticipated under the vaguest colour of lofty ideals. People want to dismiss the potential harms of deepfakes while also excitedly releasing the deepfak…

the genie is out of the bottle. the sooner you let go of the notion that it can be kvetched out of existence the less frustrated you'll feel. for example, 35 years ago PGP was a "casually malevolent" thing, enabling terrorists and pedophiles to email each other with impunity. the effort to make math illegal had (very predictably) failed and now we have encryption everywhere. did the world end? how do you feel about C…

That argument is a lot less convincing (and sincere) when we're talking about a specific tool with a very specific purpose, rather than maths itself. In much the same way, you can make pro-gun arguments based on sport shooting, but not when you're holding a landmine.

This is a very specific tool with the purpose of making it easier to lie, in an area where the kinds of lies people tell are directly injurious to society. "It's just math" doesn't fly here, and all the primary uses for such a tool are malevolent.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#215

Post by superposeur »

latexr wrote: post #182
bonoboTP wrote: post #176

Earlier quoted context omitted.

[flagged]

That’s a non-sequitur. This has nothing to do with “mainstream media”. Which, by the way, is not a singular entity you can blindly trust or distrust, don’t fall into that indoctrination trap. This is basic despotic citizen control: Get everyone to distrust those who hold them accountable; label them as irredeemable liars so you can do as you please. Edit: Removed swipes while keeping the core argument.

Hard agree. Railing against the “mainstream media” is such a cliched and, ironically, uncritical mode of argument.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#216

Post by airstrike »

mr_toad wrote: post #140
airstrike wrote: post #30

Regardless of one's opinion about this particular project, it seems obvious to me that the path forward is proving authenticity of non-AI resources rather than attempting to watermark all the AI-generated ones.

> proving authenticity of non-AI resources You’re trying to prove a negative.

Not really, I'm trying to prove "this is an actual photo from some specific certified hardware."

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#217

Post by cortesoft »

site-packages1 wrote: post #2

I don't know I really like the definitive indicator that something is AI so I can completely ignore anything else that comes from them.

I feel like it is even worse if we have a marker that is ALMOST always present on AI images. It would make us more likely to be fooled by an AI image that had them removed, because we would trust the marker.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#218

Post by bawolff »

blablabla123 wrote: post #195
bawolff wrote: post #181

Earlier quoted context omitted.

I disagree. Its mainly about having technical control and freedom. Reverse engineering how things work feels like peak hacker ethos. You don't have control of something if you can't remove it. I think ethical considerations were always a bit secondary to technical power when it came to so called "hacker ethos". After all, instructions on how to remove watermarks definitely feels like the sort of thing that would have…

There's a thing called Hacker ethic which used to be referenced quite frequently in the past: https://en.wikipedia.org/wiki/Hacker_ethic Probably it's worth reminding of also considering we're on HN here... ;)

Well, to quote from the article

> Hackers believe that essential lessons can be learned about the systems—about the world—from taking things apart, seeing how they work, and using this knowledge to create new and more interesting things.

> Access to computers—and anything which might teach you something about the way the world works—should be unlimited and total.

> Mistrust authority—promote decentralization

> All information should be free

I phrased it a bit differently, and perhaps a little less sympathetically, but i think i was more or less saying the same thing.

In any case a tool like the article that strips watermarks seems exactly the sort of thing that would fit into what i quoted above. Its mistrusting authority - there is nothing more central authority then having a literal central authority adding hard to remove digital signatures to images. It promotes freedom of information - it supports explaining how watermarks work and what they are. Its fundamentally taking apart a system, which teaches us how the system works.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#219

Post by KaiserPro »

amanaplanacanal wrote: post #202
KaiserPro wrote: post #191

Earlier quoted context omitted.

> suppose someone is using AI to modify their appearance because they're being unjustly targeted by an oppressive government Then use a mask like everyone else. digital mask, one that obscures. which is my main point, no, there isn't a legitimate need. realtime avatars don't generally have invisible watermarks, also they are running from your machine, otherwise you've got a (normally credit card) trail to your front…

As someone else pointed out: if watermarks are required, then everybody will assume an image without a watermark is the honest truth, which is obviously not true. Someone will end up in prison because of some image. This is bad.

but right now, we are eroding trust at an industrial scale.

There are no reliable tools for the end user, normal person, to work out if an image is AI or not. This erodes trust and lets bad actors get away with "oh thats AI generated" or use AI to defraud users.

Re: Remove-AI-Watermarks – CLI and library for removing AI watermarks from images

#220

Post by nancyminusone »

j2kun wrote: post #12
akersten wrote: post #7

There's an underappreciated comment in the other thread about SynthID and OpenAI [0] that captures what (IMO) the hacker ethos on this should be. We care about privacy, we should not accept tools that barcode our every digital move. (note that the counter of "well, they don't do that yet " is not particularly convincing) [0]: https://news.ycombinator.com/item?id=48200060

Building a tool that tries (and probably fails) to remove the watermark (due to the arms race that large corporate machines will win) is tacitly accepting the barcode. The hacker ethos should be, first and foremost, to run open source models locally without relying on a corporation.

Or you can just take a picture of your monitor with your phone.
Post reply on HN