[Return]

Report a post

Preview
Encryption ideas for hiding in plain sight (a ramble)

I'm fascinated by the idea of hiding messages in regular data on the open internet. Basically, my fear is that as governments begin locking down the internet more and more, it is a possibility that traditional privacy protection measures that are currently viewed with suspicion will be outright criminalized. I can already imagine it being sold as "reducing the amount of junk data on the web." This is how the regulation of encryption is characterized for amateur radio, and radio enthusiasts that are not privacy conscious absolutely eat it up.

The basic idea is to embed your message within another, much larger, message. Consider a coded newspaper, where the first letter of each line spells out a message for those that know to look for it. Now, on a computer, things are much easier. Rather than having to fill in the gaps of your message with readable text, you could instead embed your message into any data. For instance, you could spell out your message using the least significant bit of the red component of each pixel of an image, without greatly affecting the appearance of the starting image. You could also devise much more clever schemes to distribute your message throughout the medium, and I imagine this has already been done (for instance, some sort of nonrepeating pattern with all sorts of magical math stuff).

This is still not ideal, however. There is some amount of footprint being left behind. Either you are taking images that you found on the web and altering them in a detectable way, or you are providing your own, new images. Not only are both of these potentially identifiable, they also require some level of effort. Remember, the point is to be able to post these images on the clearnet, so ideally it's not a random mishmash of suspiciously subtly doctored images that practically scream "I'm hiding something, please get a warrant and start trying to crack my encryption, also all of my activity on this account can be traced to my IP address!"

However, thanks to the rise in popularity of generative AI, there is now an easy source of socially acceptable junk data. You can take a short message, quickly wrap it up in megabytes worth of junk data, and still be able to post it anywhere without raising suspicion. It is now relatively normal to run an account that only posts AI generated media, which is absolutely perfect for this. You can take an encrypted message, distribute it throughout an empty canvas, and fill in the gaps with an image generator to create a funny meme, or whatever. This can obviously be done with anything else that is popular to use AI for, too.

Of course, this AI stuff can all be done in addition to doctoring existing images, audio streams, and whatever else, the point is the ease with which you can achieve volume. So now every meme page is potentially a private message board, and every 4 hour "rain sounds for sleep" collection is potentially a pirated videogame.

You can also use AI to generate a large amount of content in which your "message" is simply junk data. So even if your embedding process leaves a footprint, you could only embed a "useful" message in every 100th post. There's even potential here, although this seems much harder, to cast doubt on whether or not someone is even using this encryption. Basically, if you could fabricate a key for which at least two posts made by the same politician decrypted to legible messages, at least one of them incriminating in some way, you would definitely get that guy to start thinking about whether you should be able to prosecute someone over the suspicion of embedding encrypted messages.
Post number No.161358
Board Off-Topic@Heyuri
Optional. Describe what's wrong with it.