PhotoDNA

PhotoDNA is a technology developed by Microsoft and improved by Hany Farid professor at Dartmouth College that computes hash values of images, video and audio files to identify similar images. PhotoDNA is primarily used in the identification of child pornography (more properly identified as Child Sexual Abuse Material or CSAM), and works by computing a unique hash that represents the image. This hash is computed such that it is resistant to alterations in the image, including resizing and minor color alterations. It works by converting the image to black and white, resizing it, breaking it into a grid, and looking at intensity gradients or edges. This technology is for certain implemented in all of your devices. If your device is connected to the internet and is using proprietary software that is using "the cloud" to function, it definitely is scanning your images and checking them against a database. How long until this gets used for stuff other than CSAM? What are your thoughts on this dystopian technology? I'm certain most Android phones scan your images by default, the only way you could circumvent it is by installing a custom open source ROM paired with open source software with no network connectivity.

Don't use proprietary software that is using "the cloud" to function

Another dystopian hell tech from bill gates, who would have thought. No doubt just like anything else they claimed was to protect us it will put in anything that is threatening to the establishment into their databases or whatever. I suppose a custom ram can stop this, only problem is most phones are not rootable, not to mention who roots their phone nowdays anyhow.

>This technology is for certain implemented in all of your devices. If your device is connected to the internet and is using proprietary software that is using "the cloud" to function, it definitely is scanning your images and checking them against a database. It's not implemented on devices. I'm pretty sure the fingerprinting code is a secret lest pedos test against it and slightly modify images so they don't match. It's implemented on the backend of large web/cloud services. Supposedly Facebook and Google use this to match against every image they might ever distribute. They've been doing this for almost a decade. >How long until this gets used for stuff other than CSAM? That started happening years ago. It's not a secret. Where this is going to get ugly is the EARN IT Act. I predict that this will be required of all services that want legal immunity for user generated content. Such a requirement is incompatible with E2EE and any other strong privacy measures.

Great stuff, hope we can use it to trace images back to the original creators. Doubt it will be used for more aggressive or mainstream scenarios, doesn't sound worth it for MS in my opinion.

>>32209 >How long until this gets used for stuff other than CSAM It almost certainly already is. The various leaks have made it exceptionally clear that the intelligence agencies are already several steps ahead of what people suspect when it comes to mass surveillance and most consumer devices are incredibly aggressive with automatically uploading all of your images to the cloud. Since they're already scanning it and probably piping it through NSA/GCHQ datacentres where they will probably also be scanning for images that suggest dissident activity (or potentially compromising images of flagged potential dissidents). But by keeping the focus on its good uses like against child porn they have an additional shield for this activity alongside the standard tinfoil hat accusations and it won't be until after its been running for 10-20 years that someone leaks it that it will become widely accepted as occurring and when it does people will just shrug and say so what.

>>32209 >How long until this gets used for stuff other than CSAM? https://about.fb.com/news/2016/12/partnering-to-help-curb-spread-of-online-terrorist-content/ About -4 years. Mission creep for this type of technology is unavoidable. It's too bad because in the long run this is going to hurt their CSAM detection abilities when the backlash hits.

>>32237 Lookup GIFCT, it's an extension of what you posted. EFF has some choice words on it.

>>32209 >android You're being schizo but there's already a thing with DRM where if you connect to the network over certain countries, your MS laptop and Android OEM phone would delete pirate copyright content. Seems like the tech is already here. >photodna code modified and used against dissenters or journalists I wish they open sourced this kind of thing so it's easier to source imageboard content. Yes, it can become malicious enough to target dissenters and journalists but that's going too far and I don't think they'll just hand over the codes to someone for free. Both targets would probably go for distros designed to avoid such scenario, and also avoid implanting of such thing (be it csam or illegal manuals) over your device which malicious actors will use to frame journalists or similar e.g, Assange.

What's next, they will ban cartoons and cosplay? This can't be happening!

>>32211 >>32250 I hope this PhotoDNA dies.

>>32255 These things will never die, and will only improve. We need laws, and law enforcement (NSA Snowden leaks).

This is an extremely old technology here's a blogpost going back as far as 2008 explaining something similar https://www.hackerfactor.com/blog/index.php?/archives/210-Bad-Science-How-Not-To-Do-Image-Analysis.html you can check his site for more info

It's nothing more than a perceptual hash algorithm. Those algorithms have been around for decades. See PHash for an open implementation. I have no idea why they market it as a panacea to CSAM. Probably to get funding from governments and to make it look like a noble project. And yes, it certainly will be used for other use cases as well, such as enforcement of copyright claims, detection of political content (after "fake news" has been outlawed)

>>32250 They can't just ban art. Art is freedom of expression. This is how uprising start and how wars begin.

>>32250 They'll probably just ban drawings with fictional lolis which isn't exactly illegal in the US as of now if PhotoDNA becomes huge.

>>32517 No war will be started because of this. Dream on.

Honestly CP and moral panicking over online is predation is not really that much of an issue. Normal users should stop being so paranoid about it. Half of if isn't even 'CP' but barely legal teens in their 16s and 17s. And what percentage of people do you think even bothers to take the legal risk to store CP at all? And no, I'm not a pedo.

>>32212 Yes, OP is missing an important dimension of what's going on here. The law mandates all websites in US jurisdiction to report CSAM material and keep records about it: https://www.law.cornell.edu/uscode/text/18/2258A A tactic when a website is disliked is for activists to upload a bunch of child porn to it, and then report it for not properly following this law. This happened to the website Gab, who were also denied access to the PhotoDNA API, which they actually wanted as it would make reporting easier. So the most immediate effect of things like PhotoDNA and the enforcement of the above law is to make it harder to create new web platforms. Once you hit the limits of manual moderation, now you have to spend time developing your own CSAM detection and reporting tools; PhotoDNA is owned and controlled by the existing corporate web platforms. The only way around this is to not be in US jurisdiction, or to not be able to see or remove user-generated content. It's still probably true that devices are being scanned somehow. Why wouldn't they? >>32241 Correct. You can read all about GIFCT on their own websites, for example how they should be removing "borderline" "terrorist" content, whatever that means: https://gifct.org/press/our-work-continues/ https://rusi.org/sites/default/files/20190703_grntt_paper_3.pdf

>>32517 I don't think theres been many wars or uprisings caused by banning art and theres certainly not one been caused by the banning of child pornography. Of all the issues with photodna "its stops me appreciating CP" is not one of them.

>>33847 Child sex abuse and child porn is not a "moral panic". It is a constant endurance that has been plaguing society since its beginning. >Half of if isn't even 'CP' but barely legal teens in their 16s and 17s. That sentence makes no sense. >And no, I'm not a pedo. Sure, buddy.

sounds like marketing bs, invariant image hashing is decade old tech. They prolly rehashed what existed under a new label and thats it.

>>32517 I think you are right but not in the way you are thinking. People who are into CSAM don't share enough of a common morality to coexist with the rest of humanity. One group must wipe the other out. There is not going to be a war. It's been a cat and mouse game for the last several hundred years as various groups find and exterminate pedos. Tech like PhotoDNA is just a response to the landscape changing. In the end there can never be any safe havens for pedos. That has some bad side-effects but most people (myself included) will give up some privacy to catch more pedos.

>>33978 He who sacrifices liberty for security will have neither.

>>33985 >He who sacrifices liberty for security will have neither. Sorry about "Achtually!"ing you. https://techcrunch.com/2014/02/14/how-the-world-butchered-benjamin-franklins-quote-on-liberty-vs-security/ The quote is much more mixed. Liberty and security is a dichotomy that should be balanced. Greater liberty is vital especially in a digital environment. But, the "liberty" to share videos and pictures of children being abused is one I think we can skip as a society.

>>33989 Pictures of kids enjoying themselves should be allowed, in all contexts. Child abuse though, ya, to the bin.

>>32209 I talked to Prof. Farid about this since he is one of my current professors. I really appreciated his justification and his background in this type of surveillance ( he has a copious amount of work in deep fakes and just surveillance), but I just couldn't agree with what he believes in for a future of internet. Look at his wikipedia page and some of the stuff he manages (Truepic as a notable example) are completely justified. His vision involves everyone adopting these types of surveillance programs in order to force 100% authenticity with any information you can get. It is true that the current age of information is producing a near equal amount of misinformation which is really evident in extreme corners. This will only get worse over time as deep fakes improve, which Prof. Farid fears as the truth will be obscured. However, With open-sourced or even perfect implementations of PhotoDNA, the freedom of expression is null with a benefit of truth and authenticity, which I can only agree to a certain degree. After pondering this and considering the fact that many of the top minds and elites are pushing this form of mass surveillance for better or worse, TOR and its related technologies seem to be a future of the free internet that we all grew up with. Only for the general population will this matter, and if a lot of this future mass surveillance is implemented somewhat justly, then its a positive for the non-tech savvy. There will always be avenues around this in the future...

>>33996 >His vision involves everyone adopting these types of surveillance programs in order to force 100% authenticity with any information you can get. Of course he has this vision. He has profitable patents on the process. (https://patents.justia.com/inventor/hany-farid). It looks like the main patent in question (https://patents.justia.com/patent/9031329) is being used in a commercial product named Truepic by way of a shell corp co-owned by Farid called Fourandsix Technologies. I can't find whether the patent has been sold outright to Truepic or is merely licensed to them, but if it's the latter, and if Truepic were to suddenly see a surge in demand, then Farid would stand to make a fortune. And let's not forget the most important part of this whole thing: who controls the database of authenticity signatures?