PureBasic Forums - English

Regretfully, I do not have time to convert (or use) this now, but this seems like the kind of tip that has been sought after and/or enjoyed here.

The premise is to encode a message into 0's and 1's, and insert zero-width characters (i.e. non displaying) into text.

https://hackaday.com/2018/04/15/hide-se ... haracters/

It seems to work fine and it looks like PB leaves those zero-width characters untouched.

The newest are special fonts, into which smallest changes are inserted.
This method is extremely robust and also works on paper prints.
Without the appropriate OCR and font, the data cannot be extracted.

The use of invisible characters offers virtually no protection and is immediately noticeable in a text analysis.
The possibilities are also so severely limited that it is no more suitable than for theoretical consideration.

are you talking about FontCode? https://spectrum.ieee.org/tech-talk/com ... plain-text

Yep, correct !

FontCode is just a Bacon cipher. Not sure why it's suddenly "news" at the moment, since Bacon's cipher has existed since 1605.

The approach is completely different
You can evaluate a bacon cipher with a table, a pencil and a sheet of paper.
For the new approach you need the font and an AI

walbus wrote:For the new approach you need the font and an AI

That's only because this updated Bacon concept uses tiny font variances instead of obvious differences. Won't be hard to crack. Just needs an OCR app to look at every character and see if they differ, even by one pixel. If they do, then it's obvious a Bacon cipher in use.

Note: These are not just my thoughts; other people on the internet have echoed what I said.

You guys think it's too trivial.
The first thing is to hide it from the human eye.
Then it's about transferring the message from a to b unnoticed.
You also cannot analyze such text with any OCR app.

Such a message will always be additionally (AES) encrypted.
A change refers then to the complete text, not to a part of a text.

It can then no longer be proven whether a text has hidden content.
You can assume it after a special analysis, but you can never prove it.
You need a special analysis tool to determine if there are any conspicuities in the text.
But your analysis will never yield more than noise.

Without the corresponding special software it is impossible to decrypt this.

If people develop something like that with a lot of effort,
it's not expected to be useless three hundred year old know-how,
that would make absolutely no sense, I guess.