Steganography

From KTN AI Forensic Group

Here we're talking about hiding data within data so it appears to be something innocent. Here are a couple of trivial examples:

Picture files

Supposing you wanted to send some text through unencrypted email but didn't want a keyword scanner to pick it up. One way we've all seen of doing this is to convert the text into a graphic image and send that instead. Spammers use this technique a lot to avoid text filters finding words like viagra.

There are far more advanced techniques for storing data inside a graphic image such that it can only be seen and extracted if you know it's there - perhaps this article can be expanded?

Protocols

Another trivial example - off the top of my head so it may not be practical. Supposing you wanted to send messages between two points without arousing suspicion. Do this by sending normal messages between he two sites using a standard messaging protocol BUT encode the actual message in the name of the recipient instead of the body of the message, leaving the body of the message as innocent chatter.

Real-world protocol steganography can use fields in protocol headers to encode information - false error codes and so on - making a hidden data channel very difficult to detect.

Application of AI

It's hard to spot steganography taking place. This is the whole point of it. The data is in plain view, hidden in a huge volume of innocent data. On the face of it, AI stands a better chance of spotting anomalies as no human can examine a huge volume of data.

Please expand this article if you can!

--Frank 09:25, 13 February 2009 (UTC)