FR version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
61% Positive
Analyzed from 4520 words in the discussion.
Trending Topics
#watermarks#watermark#image#trust#should#something#don#tool#generated#images
Discussion (237 Comments)Read Original on HackerNews
[0]: https://news.ycombinator.com/item?id=48200060
Much like how the entirety of Hollywood, book publishers, academic publishers, and game developers have won against piracy despite being some of the largest corps on earth and dedicating untold billions to the issue over the past 30 years?
Hobbyists do not. ROI is not a factor.
Losing to piracy would see companies like Netflix and Spotify not thriving.
I don't really see it. I think it's important to win on both fronts.
> works for Google
Gee, I wonder why...
A knife and a handgun aren’t comparable to a machine gun and a bomb. When you have equal access to all of those, the damage you can enact is exponentiated.
You could lie with text before, but it took effort and time and skill to do it convincingly. You could also lie with images but they took even more time and effort and skill, greatly limiting the pool of people who could do it and the possible damage.
When anyone anywhere can convincingly lie and have it do two laps around the world in a matter of minutes, the whole game changes.
It’s becoming very hard to believe that people making arguments like yours are doing so in good faith. Maybe you’re not even a person but a shill bot. That’s a very real and trivial possibility today, which is the whole point and illustrates the problem.
- Watermarks are optional by AI provider so bad actors will circumvent by using another provider
- GH project proves watermarks can be removed
Given these, trying to ensure "truth" is a futile effort unfortunately, and watermarking only gives companies advantage to violate privacy
Now Nancy, a tech-phobic waitress who has a grudge against her coworker can make up an entire scenario with one prompt and her colleagues might blindly believe her.
Let's not pretend they're the same thing.
Gen AI is inevitable. Watermarking is likely futile. But in my opinion it is still very important to discuss how, as a society, we're going to live in a post-truth world now that anybody can, IN SECONDS, not only fabricate a story but also spread it to thousands of people through their social media.
Do you want to make it easier for the next Stalin?
I'm not entirely sure how hiding that something is GenAI fits in here. It surely doesn't have anything to do with Privacy though.
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 been in phrack back in the day.
Probably it's worth reminding of also considering we're on HN here... ;)
but I also live in a society that requires trust to function. making a tool the obliterates that trust(genAI imagery pipelines) then creating a tool that makes it trivial for normal people to remove any hint of controls over said trust eroding system is, toxic.
I get the argument about not putting in fingerprints that identify users, Good I agree. But this also removes the things that identify this as an AI image.
Now, what are the legitimate uses of that?
No really, why would I _need_ to remove a watermark for _legitimate_ purposes? Assuming that watermark is generic, rather than a fingerprint of a specific person
When removing the watermark is easy, a very legitimate purpose of making the code to do it publicly available is to make a public demonstration that it's easy to do.
As for content use cases, suppose someone is using AI to modify their appearance because they're being unjustly targeted by an oppressive government. That government naturally bans doing that because they want to be able to identify and arrest their critics, so now if you make videos with your real face you get arrested but if you use a generated avatar then the watermark enables automated censorship because the government orders anything with the watermark to have its reach automatically restricted.
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 door. plus a video stream
also if you are generating stuff from a public provider, then tracing people isn't that hard to do.
Without truth freedom and privacy are endangered too.
The other comment talks about laws that can already handle that. How if images, video and audio aren’t reliable proof anymore?
Also note that people have been falling for obviously watermarked videos already.
And even if they weren't, wouldn't that just make them more gullible towards non-watermarked models?
ie the video of garbage being thrown out the windows that his team already confirmed was real:
https://www.kptv.com/2025/09/03/trump-says-video-showing-ite...
Also the Lincoln Project video footage him him stumbling while walking and over his words: https://www.forbes.com/sites/mattnovak/2023/12/04/donald-tru...
It just defaults to 1K. But I didn't see anything in the docs stating that it's just a simple upscale for larger resolutions.
https://ai.google.dev/gemini-api/docs/image-generation#gener...
> Produce production-ready assets with native 1K output and built-in upscaling to 2K and 4K resolutions
The API doc you linked is misleading.
Qwen-Image-2512 / Z-Image / Flux.2 absolutely crush SDXL if you're actually generating moderately complex scenes.
Sorry, how does using AI to generate images have anything to do with this? Image generators cannot insert watermarks into things they did not generate, and it seems highly unlikely that you will get a false-positive watermark on human-generated art, especially if, as the readme says, these watermarks have high enough fidelity to trace to a specific session id. Plus the modifications to the image needed to erase watermarks would necessarily change the thing being "preserved."
[edit]: the more I read the more I'm convinced, the claimed use cases in the README are bullshit and the real reason is to provide a tool that helps people bypass "AI-generated" labels on social media for AI slop.
> Image generators cannot insert watermarks into things they did not generate
It's actually very easy to take a real image, ask Gemini/ChatGPT to modify some tiny part of it (could be something as silly as lighting/shadow/etc), and often the resulting image will be detected by their watermarking tools. This way you can easily present any real image as AI-generated.
If you take a photograph that is misidentified as AI generated, you can “preserve the historical record“ by using this tool before publishing the image.
(Anyone know the false positive rate with watermark IDs, would’ve hoped it’s like zero)
We need to get to a place where we're checking digital signatures to see which humans were involved in the creation of something, and ignoring everything else.
The takeaway should be that watermarks aren't very reliable and you should keep your guard up anyway.
I'd be more interested in some kind of trusted Non-AI watermark.
This is something that could get integrated into cameras for example. However, considering how much AI-processing we already have in "normal" photos, it will be difficult to decide where to draw the line.
But then again, the photo itself would be authentic in this case strictly speaking.
It's not an easy problem. If I had a good solution, I would have tried to monetize it already ;)
Trust should be between people, not a person and some data.
So showing how easy to generate "false" data, this makes it more obvious for people focus on other people. Trusting people makes life much easier in my experience, while focusing on data, again in my experience, is a game of cat and mouse.
Sure, but how do you apply that to a society at large where powerful people are interested in making everybody distrust all reliable sources of information?
AI watermarks are no panacea, but at least they are a clear signal of what not to trust.
Isn't that the scenario the watermarks are useless against? Adversarial governments or anyone with enough money will be the ones who can generate images without watermarks even if you force them on the proles.
> AI watermarks are no panacea, but at least they are a clear signal of what not to trust.
Which seems like it only makes the actual problem worse? If most of them have watermarks, that only encourages people to put more trust in the ones that don't, even though those are the ones "powerful people" can still forge to manipulate everyone. What good is something that increases the credibility of adversarial government forgeries?
Only people with resources will be allowed to make content that is AI generated passed off as real.
Pandora's box is open. Instead of making a multi-tiered privileged society, we need to fundamentally restructure society to adapt.
Before that restructuring occurs it is critical to keep the playing field level. These are not tools that should be controlled by a minority authority, they are far too dangerous.
You can’t, it’s an inherent contradiction. Human social structures have sophisticated and robust evolved mechanisms for establishing and maintaining trust. These dynamics are not one option among many, they are the optimum. By their definition they don’t scale to strangers around the planet. This is an immutable factor in why we have spam, bank fraud, etc. We want the benefits of trust without the cost of local constraints but wishing doesn’t make it so.
We now have the tools to increase trust in specific information, for example: by signing images that need high trust for things like news reporting using camera hardware root of trust with time and geo stamping. If signatures are removed, that's back to a default low-trust state.
That is how we ended up with the situation where "reputed" media organizations peddle daily lies or selective truths that are useful to their benefactors.
It's too late already. We live in a post-trust society now.
Preserving the illusion that that assumption is still useful only helps the people who are exploiting it.
would you say the same thing about a ROT13 breaker, or would you recognize how laughably naive that sounds?
AI generated images have never existed before, they will break our ability to use the digital tools we have built society on if we let them. Ensuring they can be identified and have attribution tracing data embedded is a reasonable step to prevent abuse.
That said, this tool is incredibly lossy, garbling text, completely changing shapes. It also fails to remove the new gemini spark / placement that moved since yesterday.
I know of at least one music GenAI service whose ToS forbid me from using their productions in ways that are incompatible with my local rights. If Google decides, as they did with YouTube, that they'll enforce this company's watermark even though I have a right to use their results then I've passively accepted that a (foreign) company decides which of my legal rights I can exercise.
For a more pessimistic outlook: for the GenAI companies that work for the military, are those pictures also watermarked? Because if watermarks only apply to one group then the "implicit trust" argument doesn't work.
I'm in favor of watermarking the same way I am in favor of paying artists for their work. But just as with youtube-dl and DVD decryption libraries, tools like these are necessary to level the playing field.
Making untraceable assets means parties can’t be held liable for harms.
Just because a bad actor might poison a toddler doesn’t mean you should sell them the arsenic.
Just because the effects of arsenic on the neurons of a brain are interesting means you should feed them to a toddler and watch the effects unfold.
I support the legal freedom to pursue any idea, but we should also mentor our colleagues about the consequences of our projects and avoid unnecessary harm.
All watermarks are cures that are worse than the diseases they are trying to prevent (and never achieved to even cure anything).
Watermark is a blatant violation of legitimate privacy expectations, while not preventing anything because they can be easily removed.
So all the honest usages are punished, while illegitimate ones are not caught.
People want to dismiss the potential harms of deepfakes while also excitedly releasing the deepfake-hider-3000 and saying they just really, really care about privacy (for people who make deepfakes).
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 Chat Control and numerous other initiatives to roll it all back?
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.
as for the rest of it: would banning a tool that breaks ROT13 make ROT13-encrypted communications safe?
You’re trying to prove a negative.
putting cyphertext in high frequency noise is old news. in generative land would be far more interesting to use the generative flexibility to encode in macrostructure.
Then they ask us to do it by hand?!
No, there's no law requiring disclosure of AI use yet, as far as I know.
> pretending you're not American when asked
Many people do so for various reasons, for example saying you're Canadian when on vacation. Do you count that as evidence that they're "ashamed" of being american? Or that being american is automatically a bad thing, because sometimes people hide it?
https://futurism.com/artificial-intelligence/real-monet-ai-c...
Should people be scared of their food containing Red Dye 3? Yes. Aspartame? Maybe. MSG? Probably not. Dihydrogen Monoxide? Nope.
Should you feel "shame" for using AI to generate fake news? Yes. Art? Not really. See the numerous examples of people disliking famous human-made art just because it was presented as AI.
- Rocky